[HIRING] Technical AI Automation Specialist — Long-Term Team Collaboration

Jobs
1

[HIRING] Technical AI Automation Specialist — Long-Term Team Collaboration

Hi everyone,

We’re building a new AI Automation Agency focused on helping businesses automate customer communication, lead handling, appointment booking, and repetitive business processes using AI.

We are currently building the core team and looking for a serious, reliable Technical AI Automation Specialist who is interested in a long-term working relationship.

This is NOT a one-off project and we are NOT looking for someone to simply complete a task and leave.

We’re looking for someone who wants to be part of the technical side of an early-stage agency and work with us as real client projects come in.

What you would work on

• AI Voice Agents / AI Receptionists
• Retell AI / Vapi or similar platforms
• n8n automation
• REST APIs & Webhooks
• CRM integrations
• Calendar integrations
• Twilio / telephony
• OpenAI / LLM APIs
• Lead qualification and routing
• SMS / Email automation
• Data handling / databases
• Deployment and configuration
• Testing and troubleshooting
• Monitoring and maintenance

Example system

A typical system could look like:

Incoming call
→ AI Voice Agent
→ Understand caller’s intent
→ Collect name / phone / service
→ Qualify the lead
→ Book appointment
→ Update CRM
→ Send notification
→ Transfer urgent calls to a human

We need someone who can understand and build the complete system, not just one isolated component.

What we are looking for

You do NOT need to know every tool listed above.

What matters most is practical experience and the ability to connect different systems into a reliable production workflow.

Strong experience with several of the following is highly valuable:

• n8n
• Retell / Vapi
• Twilio
• APIs / Webhooks
• CRM integrations
• Calendar integrations
• OpenAI / LLM APIs
• Databases
• Git
• Deployment
• Debugging / troubleshooting

We especially care about production thinking:

Build → Test → Deploy → Monitor → Maintain

We want someone who understands things such as:

• API failures
• Error handling
• Edge cases
• Retries
• Data validation
• Human handoff
• Logging
• Reliability
• Monitoring

Important — this is an early-stage team

We are a newly launched agency and are currently building both the team and client pipeline.

There is no guaranteed fixed monthly workload at this stage.

We are NOT asking anyone to work for free.

Any ongoing work, project, compensation, and working arrangement will be clearly agreed upon before the work begins.

Our goal is to build a long-term relationship with the right person as the agency grows.

To apply

Please reply with:

  1. A short introduction

  2. Your strongest technical skills

  3. Your experience with n8n

  4. Your experience with Retell / Vapi / Twilio

  5. 2–3 AI automation projects you personally built

  6. Links to demos / Loom / GitHub / screenshots if available

  7. What you personally handled in those projects

  8. Your expected compensation / rate

  9. Your preferred working model

  10. Your availability

  11. Your timezone

  12. Roughly how long you would need to build a production-ready AI receptionist similar to the example above

Please focus on what YOU personally built, rather than only describing your company’s or team’s work.

We are looking for someone serious, reliable, communicative, and interested in building something long-term.

Generic applications will not be prioritized.

7 « J'aime »
2

Hi Ahmed — before I send a full application, one quick question: do you already have any signed or funded client projects that the technical specialist would be able to work on soon, or is the agency currently still building the initial client pipeline?

Ahmed Refaei <notifications@n8n.discoursemail.com>于2026年9月5日 周六22:43写道：

3

Hi Ahmed,

I’m Oyekola, an early-career AI automation/workflow engineer in Lagos (WAT, UTC+1). My strongest areas are n8n, REST APIs/webhooks, structured AI routing, Airtable/Google Workspace integrations, state handling, validation, debugging, and documented handoff. I also use Python/JavaScript/SQL where needed.

My strongest direct n8n proof is FlowForge AVA: a portfolio build for WhatsApp-style intake, validation, intent routing, specialist agents, Airtable CRM actions, calendar checks, and human handoff. I personally built the workflow and its state/routing logic. It is verified portfolio work, not a production client deployment: https://github.com/oyekola-ololade/FlowForge-Ava-AI

Other work I personally built includes NewsIQ (Python/SQL + five n8n workflows; partial MVP) and MailIQ (substantial historical email-intelligence prototype, currently offline/under repair). GitHub: https://github.com/oyekola-ololade

I don’t yet have hands-on Retell/Vapi/Twilio production experience, so I wouldn’t represent myself as ready to own the voice/telephony layer alone today. I would be strongest initially on the n8n/API/CRM/calendar/data/reliability side, and I’m open to a paid trial on that slice while learning the voice layer.

For a full production-ready AI receptionist including an unfamiliar voice stack, I’d want to inspect the exact CRM, telephony, booking rules, failure paths and acceptance criteria before giving a defensible estimate rather than guess.

Rate: open to a small fixed paid trial first; for ongoing work I’m flexible based on scope and responsibility.
Working model: remote project/contract, long-term if fit is good.
Availability: can start immediately; roughly 20–40 hrs/week.
Timezone: WAT (UTC+1).

If useful, give me one bounded n8n/API/CRM workflow or repair task and I can show you how I structure, test and document it.

1 « J'aime »
4

Hi Ahmed,

This is very close to the kind of systems I build.

I’m a Full-Stack / AI Automation developer working mainly with n8n, APIs, webhooks, LLMs, Next.js, Supabase, Python and JavaScript.

My focus is not just connecting nodes, but building complete systems around the automation layer.

One relevant example is MadaVoyage, a travel platform I built with Next.js that includes a live AI voice agent using ElevenLabs Conversational AI. Visitors can talk directly with the assistant to ask questions about trips and the booking process.

You can test it here:

I’ve also built n8n workflows for lead qualification, email processing/classification, automated responses, document analysis/RAG and API-based business workflows.

I haven’t worked extensively with Retell/Vapi/Twilio in production yet, but API integration and orchestration are a core part of my work, so connecting the voice layer with n8n, CRM, booking systems and business logic is exactly the kind of project I’m interested in.

Portfolio:

I’m based in Madagascar (UTC+3), available remotely and interested in a long-term collaboration.

Happy to discuss or start with a small paid technical task.

5

Hi Ahmed

1. Short introduction
My name is Henry. I’m currently learning and building with n8n and AI automation, and I have a background in JavaScript/web development. My main interest is using n8n, AI, APIs and JavaScript to build systems that automate repetitive business processes, particularly around leads, customer support and sales.

2. My strongest technical skills

  • JavaScript
  • n8n workflow automation
  • REST APIs
  • Webhooks
  • JSON/data transformation
  • Working with LLMs/AI
  • Building AI agents/workflows
  • Basic RAG concepts and implementation
  • Connecting different services through APIs

3. My experience with n8n
I’ve completed the n8n Academy courses and have been putting what I’ve learned into practice by building projects.

I’m comfortable with things like workflows, triggers, expressions, data transformation, HTTP requests, webhooks, credentials, APIs and JavaScript/code nodes.

4. Retell / Vapi / Twilio
I haven’t worked extensively with Retell, Vapi or Twilio yet. I’ve studied how these types of tools can be used in AI automation, This is actually one of the areas I’m very interested in learning.

5. AI automation projects I’ve built

I’ve been building projects around:

  • RAG AI Agent – working with knowledge retrieval and AI responses.
  • AI Customer Support Workflow – an automation designed to receive customer questions and generate relevant responses using AI.
  • Lead Automation / AI Sales Workflow – working on automating lead handling, qualification and follow-up.

6. Project links
sales pipeline
RAG for real estate agency

7. What I personally handled
These are personal projects, so I handled the workflow design, n8n implementation, API connections, AI logic, data handling and testing myself.

8. Expected compensation
At this stage, compensation isn’t my main motivation.

I’m primarily interested in the learning opportunity, real-world experience and the chance to contribute to a serious team.

9. Preferred working model
Remote, and I’m comfortable working collaboratively with a team.

10. Availability
I’m available and committed to putting serious time into this. I’m looking for something where I can consistently contribute and learn.

11. Timezone
WAT (West Africa Time), UTC+1.

12. Estimated time to build an AI receptionist
I want to be honest here because I haven’t built a production-ready AI receptionist before.

For a system similar to the example, I would need time to properly understand the requirements, learn the voice stack being used, build the workflow, integrate the necessary services, test different scenarios and handle edge cases.

I’m applying because I’m serious about learning this properly, I’m willing to put in the work, and I’d really value the opportunity to learn alongside people who are already doing this at a higher level.

I’m comfortable starting from where I am, taking feedback, doing the work and improving quickly.

Thank you for considering me. :folded_hands:

6

Hi Ahmed,

I’m an AI automation engineer from Karachi, 2 years with n8n and AI agents. I run Automate Solutions and also work as an AI Developer at Saylani Tech.

Top skills: n8n, Python, FastAPI, LangChain, LangGraph, RAG, OpenAI/Claude APIs, Docker.

n8n experience:
My main tool. Built lead intake flows, CRM and calendar syncs, email/SMS automations, with retries and error handling so nothing fails silently.

Retell/Vapi/Twilio:
No direct use yet, being honest here. I know the parts around it well (intent parsing, lead scoring, CRM updates, human handoff) and pick up new APIs fast.

Projects I built:

  1. Job discovery bot in n8n. Pulls listings, scores them with AI, saves to Sheets. Runs on its own.
  2. Image pipeline for a healthcare brand. Handles retries and bad data.
  3. Shirt on model image flow for a DTC brand. Error handling built in.

Can share Loom or repo links if needed. I built and owned all of these myself, start to finish.

Working model:
Long term, open to a small paid test task first.

Availability:
Flexible, can grow hours as work comes in.

Timezone:
PKT (UTC+5), overlaps well with Gulf hours.

Timeline for the receptionist example:
2 to 3 weeks for v1, while I pick up Retell/Vapi. Faster after the first build.

Happy to hop on a call.

7

Hi — this looks closely aligned with the systems I’ve been building.

My focus is Revenue Systems and AI/automation workflows, with hands-on work across n8n, HubSpot, Salesforce, REST APIs/webhooks, PostgreSQL/Supabase, and AI-assisted workflows.

Recent implementations include:
• Product/signup signals → qualification → CRM routing and sales handoff
• API/webhook integrations with identity resolution and duplicate controls
• AI-assisted account research and prioritization
• Production-oriented n8n workflows with retries, failure handling, logging, and human-in-the-loop paths

I’m particularly interested in the CRM, lead qualification/routing, API integration, and AI orchestration side of your client work.

I haven’t yet used Retell/Vapi/Twilio extensively in production, so I wouldn’t overstate that experience, but I’m comfortable working with APIs and learning the platform-specific layer quickly.

Portfolio / implementation examples:

Happy to start with a well-defined paid task and see if there’s a good fit for longer-term collaboration.

8

Hi Ahmed — your example flow (call → intent → collect → qualify → book → CRM → notify → handoff) is mostly an orchestration and reliability problem; the voice layer is the thin edge of it. Answering your list in order, and being straight about the gaps.

  1. Intro: Koray Nar, Türkiye (UTC+3). Automation developer working daily with Python + Claude agents. I run scheduled autonomous agents with approval gates for my own lead-hunting and outreach, so “build → test → deploy → monitor → maintain” is my normal week, not a slogan.

  2. Strongest skills: Python, Claude/LLM APIs (strict JSON outputs, tool use), REST APIs and webhooks, n8n (Code / IF / HTTP / file nodes), MCP servers, Git, tests. Habits: input validation, retries and idempotency, dedupe state, logging, explicit human-handoff paths.

  3. n8n: github.com/koraynar/n8n-ai-workflows — three workflows built and executed headless through the n8n CLI: webhook lead intake with scoring and routing (hot → “call within 2h” + callback reply; cold → nurture), Claude email triage with JSON-verdict parsing, and a scheduled feed scan with state-file dedupe. Real execution output is committed. No client n8n instance in production yet — the README says so plainly.

  4. Retell / Vapi / Twilio: no production build. They use the same webhook/API pattern I work with daily, but I’d want a paid ramp week on telephony (numbers, call events, A2P compliance) before calling any voice system production-ready.

5–7. Projects I personally built, solo, end to end:

  • Lead-intake assistant demo for Turkish service businesses (WhatsApp-style): instant reply → qualify → book, with a live-filling business panel — koraynar.github.io/aninda-asistan (labeled as a simulation).
  • doc-extract-mcp — MCP server, 6 document tools, path-confined, 42 tests — github.com/koraynar/doc-extract-mcp
  • trendyol-seller-mcp — marketplace seller-API MCP, env-gated writes, PII redaction by default, 25 tests — github.com/koraynar/trendyol-seller-mcp

Design, code, tests and docs in all of these are mine.

  1. Rate: from $25/hr, or fixed price per scoped build (my preference for client projects).

  2. Working model: project-based first, retainer once there’s steady flow. Written scope before work starts. White-label delivery is fine.

  3. Availability: 15–20 hrs/week now, can grow with the pipeline.

  4. Timezone: UTC+3 — full EU overlap, US East mornings.

  5. AI receptionist estimate: the orchestration layer (intent → qualify → book → CRM → notify → handoff, with logging and retries) testable in 1–2 weeks. I won’t quote the voice layer until I’ve done the paid ramp week on Retell/Vapi — I’d rather give you a real number then than a guess now.

Happy to do a 20-minute screen-share of the lead-intake build before you decide anything.

9

Hi Ahmed,

Answering your questions in order, but one thing up front so it frames the rest.

I run a small AI intake company of my own, Astracore, out of Vancouver. I’m not
mentioning that as a conflict, I’m mentioning it because it’s the reason I can
answer question five with something you can go and use rather than a description.
There’s a live receptionist agent on astracore.ca, the button in the bottom right
corner. Ask it something it can’t know and see whether it makes something up. I’d
rather you judge the work than read about it.

More on how that sits alongside working with you at the end.

  1. INTRODUCTION

Jeff Bowen, Vancouver, Canada. I build and run AI voice and intake systems for
service businesses. Solo operator, so everything below is work I did myself, not
work my team did.

  1. STRONGEST TECHNICAL SKILLS

n8n to a deep level, including custom code nodes and patterns that aren’t in the
docs. Vapi assistant design, tools and system prompt engineering. REST APIs and
webhooks, including the awkward ones with no ready-made node. Calendar and CRM
integration. Data pipelines with deduplication and validation. Debugging live
systems where the failure only shows up under real traffic.

The thing I’d point at rather than a tool list: I’ve had agents in front of real
callers, which teaches you different lessons than building one does.

  1. EXPERIENCE WITH N8N

Around 89 workflows in production across intake, enrichment, outbound sequencing,
document processing and content pipelines. Self-hosted instance.

The pattern I use most: an API bridge workflow, a single webhook fronting an HTTP
Request node holding the stored credential, so every other workflow calls one
internal endpoint instead of each carrying its own auth. Makes credential
rotation a one-place change and stops secrets spreading across a dozen workflows.

I also handle the multi-tenant side, keying shared instances by client ID and
moving sensitive clients to dedicated in-region instances when data residency
demands it.

  1. EXPERIENCE WITH RETELL / VAPI / TWILIO

Vapi in production. Assistant configuration, custom tools hitting n8n webhooks,
knowledge files, voice selection, latency tuning across transcriber, model and
voice.

Twilio for numbers and conditional call forwarding. That last part is more
fiddly than it looks: it’s carrier-specific, it’s a settings change rather than a
port, and it has to be tested per client or you find out the hard way that their
provider handles busy and no-answer differently.

Not used Retell. Same shape of problem and I’d expect a short ramp, but I won’t
claim it.

  1. THREE THINGS I BUILT

a) The Astracore receptionist. Inbound call, intent, qualification, live calendar
read, booking, CRM write, confirmation to both sides, transcript and summary,
escalation to a human for anything urgent. The rule that took the longest to get
right is what it does when it doesn’t know something. It takes a message. It will
not quote a price outside its knowledge base or commit to a slot that isn’t free.
Live on astracore.ca if you want to try to break it.

b) An outbound sequencing pipeline. Source, enrich, deduplicate, segment by
signal, batch to daily send limits, push to the sending platform, then write back
so nobody gets contacted twice. Handles partial API failures mid-batch without
double-sending, which is the failure that actually costs you.

c) A document intake pipeline. OCR, field extraction, validation, and routing
into the right downstream system, with anything below a confidence threshold
diverted for a human to look at rather than written silently.

  1. LINKS

astracore.ca, live agent in the bottom right. That’s the demo. Happy to do a
screen share walking through the n8n side, since the workflows themselves aren’t
public.

  1. WHAT I PERSONALLY HANDLED

All of it. Architecture, build, integrations, prompt design, testing, deployment,
monitoring and the client-facing side. There’s nobody else.

  1. RATE

$70 USD an hour for ad hoc work.

For receptionist builds I’d rather quote per build than hourly, because it’s a
known shape of work and you shouldn’t be paying for my learning curve on your
clock. Roughly $1,200 to $1,800 depending on how many systems it touches, and a
monthly figure per live client if you want me monitoring and maintaining them
rather than handing them over.

Open to discussing all of that. I’d rather land on something that works for an
agency at your stage than quote you into a corner.

  1. PREFERRED WORKING MODEL

Contractor, per project or per build, with a maintenance retainer on anything
that stays live. I’m comfortable working under your brand and staying invisible
to your clients if that’s how you want to run it.

  1. AVAILABILITY

15 to 25 hours a week reliably, more for a defined push. I keep my own workload
running alongside, which I’d rather be upfront about than have you discover.

  1. TIMEZONE

Pacific, UTC-8. Good overlap with North America, workable with Europe in my
mornings.

  1. HOW LONG FOR A PRODUCTION-READY RECEPTIONIST LIKE YOUR EXAMPLE

One to two weeks from having the client’s information in hand. A simple one in a
few days.

That assumes the slow parts are the client’s, not mine: getting their prices,
hours, service area and qualifying questions out of them, and getting access to
their calendar, CRM and phone carrier. The build itself is maybe three days. The
rest is waiting on people and testing forwarding behaviour, which is carrier
specific and always throws up one surprise.

I wouldn’t put one live without a test call the client watches, where they see
the booking land in their own calendar and the contact appear in their own CRM.
That step catches more than any amount of internal testing.

ON RUNNING MY OWN THING

Worth being direct about it since you’ll find the site anyway.

I don’t see a conflict and I don’t think you will either. There are more service
businesses that need this than either of us will reach, and I’m not in a position
to be precious about who I build for. What I’d want is simple: no exclusivity and
no non-compete, I don’t approach your clients, you don’t approach mine. If that
works for you it’s a non-issue.

What you get out of it is someone who has already made the mistakes on his own
money rather than on your first client’s.

If a straight contractor arrangement suits you better, that’s fine and I’m in.
If at some point you’d rather I carried the whole delivery side under your brand,
that’s a conversation worth having too, but it only makes sense once there’s a
pipeline to feed it.

Where is your pipeline sitting right now? That’s the thing that decides whether
this is worth either of our time yet, and I’d rather ask than guess.

Jeff Bowen
Astracore, Vancouver
astracore.ca

10

Hi Ahmed — your checklist, in order.

Background. I build production systems: n8n orchestrating, Airtable as the single source of truth, LLMs for intent only. Before automation: technical sales for manufacturers including Legrand and Somfy, purchasing on civil-works contracts, then running the operations of an industrial site.

n8n. Daily. Code nodes, OAuth and webhook signature verification, a queue in front of webhooks, exponential retry with a dead-letter queue, idempotency keys from the event id so a redelivered webhook never sends a second message. State in the database, not in model context. Secrets in the credential store.

Monitoring. Alerting on failure is the easy half. What costs money is the run that should have happened and did not. I ship a daily completeness check: expected against actually ran, gap flagged.

Built myself:
— WhatsApp sales agent, live in production on the official Cloud API. First reply to a paid-ad lead under 6 seconds, 24/7, four languages. Voice-note transcription, booking, human handoff out of scope.
— Construction operations on Airtable: quotes → site work → invoicing → cash-flow. Agents summarise site managers’ weekly reports and propose next week’s actions. They propose; a human decides.
— Reconciliation across three systems naming twelve line items eighty-two ways: 75 human decisions in month one, zero in month two. Systems real, data synthetic — said up front.

Portfolio: https://www.upwork.com/freelancers/~015be6a6767bdd5642

Rate. €35/hour for build and maintenance, or fixed price per system once scope is clear — I prefer fixed price. First trial task at a flat €150 whatever its size, so your risk is bounded.

Casablanca, UTC+1. 30+ hours a week, available now. English, French, Arabic.

Happy to take a short call.

— Jamal Cherki Boumsik

11

I’m Seyed Naser Sajjadi, an innovation/product manager and hands-on automation builder with 15+ years of experience turning operational requirements into digital products, workflows, and technology projects. I’m currently focusing increasingly on n8n and AI automation.

My strongest areas are business-process and workflow design, APIs/webhooks, data validation, integration testing, Git, deployment coordination, and production-oriented reliability.

With n8n, I’m currently building client-facing automation demos and workflow prototypes around APIs, webhooks, structured data, and AI-assisted processes. I want to be transparent that I’m still building my n8n production portfolio rather than claiming years of production n8n experience.

I don’t yet have production experience with Retell/Vapi/Twilio, but I’m comfortable working with APIs and learning a new integration layer quickly.

Relevant systems I have personally worked on include:
• Meydano — a live sports booking platform with booking rules, payments, SMS, notifications, and operational workflows.
• SAMAP — an enterprise project-operations platform covering users/roles, projects, notifications, documents, deployment and operational processes.
• A current n8n-based business automation demo designed for presenting practical automation use cases to companies.

I personally handle product architecture, workflow definition, business rules, validation/testing, integration decisions, and delivery coordination.

LinkedIn: http://linkedin.com/in/seyed-naser-sajjadi

Compensation: I’m open to starting with a small fixed paid trial task, then agreeing on an ongoing rate based on scope and responsibility.

Working model: remote project/contract, with interest in long-term collaboration if there is a good fit.

Availability: I can start with a bounded paid trial immediately.

For the AI receptionist example, I would prefer to first inspect the CRM, booking rules, telephony stack and failure/acceptance criteria. A first orchestration MVP could reasonably be scoped in roughly 2–3 weeks, but I would not give a production commitment on the voice layer before validating Retell/Vapi/Twilio requirements.

I’d also be happy to start with one bounded n8n/API/CRM workflow so you can evaluate how I structure, test, and document the work.

12

Hi Ahmed,

I’d be interested in joining the team. My main focus is AI automation and workflow engineering with n8n and Make.com, including API integrations, webhooks, AI/LLM workflows, CRM automation, lead handling, and troubleshooting.

I’ve personally built AI automation systems involving an AI booking assistant, automated lead processing/qualification, and multi-step workflows connecting different services through APIs and webhooks.

I’m particularly interested in the long-term aspect of this rather than just taking on one-off tasks.

I’ll send you a DM with more details about my projects and experience.

13

Hi Ahmed,

I’m Paata, a Canada-based AI automation and full-stack developer working with n8n, Make, Python, JavaScript, PostgreSQL, REST APIs, webhooks, OpenAI, Claude, React and Supabase-style systems.

I recently built TenderScope, a construction intelligence SaaS platform with multi-source data ingestion, company intelligence, AI analysis, recommendations, scheduled workflows, duplicate protection, retries, logging and recovery from stale runs: https://www.tenderscope.ca/

I personally designed the data pipelines, entity resolution, AI matching, database layer and production reliability controls. I also build CRM workflows, lead qualification, automated follow-ups, API integrations and human approval paths.

I have not worked extensively with Retell or Vapi in production yet, so I would be strongest initially on the n8n, API, CRM, data and reliability side. I’m comfortable learning the voice layer when the workflow and acceptance criteria are clear.

I’m available for a long-term remote collaboration from British Columbia, Canada (Pacific Time), with 30+ hours per week available. My starting rate is $35/hour, depending on scope and responsibility.

I’d be happy to begin with a small paid trial so you can evaluate how I build, test and document production workflows.

14

Hi Ahmed,

I’m Yurii, based in Germany. I’m currently focused on AI and business-process automation using n8n/Make, OpenAI, APIs, webhooks and Google Workspace.

I’ve personally built several practical automation projects, including:

• AI lead qualification → scoring → Google Sheets → personalized email draft
• AI email classification and automated reply workflows
• Invoice/PDF processing → AI extraction → structured data in Google Sheets
• Lead/outreach workflows with automated follow-ups and CRM-style tracking

My strongest area is connecting AI with actual business workflows — lead handling, email, data processing, APIs and routing.

I don’t yet have production experience with Retell/Vapi/Twilio, so I don’t want to overstate that. But I’m comfortable with APIs/webhooks and I’m actively expanding into voice automation.

I’d be happy to start with a small paid trial task so you can evaluate my work before committing to anything long-term.

I’m based in Germany (CEST) and can work remotely alongside my current job.

For compensation, I’m flexible for the first paid trial and would prefer to agree on a fixed price based on the scope.

If useful, I can also share screenshots or a short demo of the workflows I’ve already built.

Best,
Yurii

15

Hi Ahmed,

I’m Zay, a developer specialising in business automation and voice integrations. I’m interested in taking responsibility for the technical delivery as your agency brings in clients, including the maintenance that follows launch.

My background spans more than eight years working on business systems and the operations they support. My work has involved developing internal tools, connecting data and services, troubleshooting failures and making systems usable by the people who depend on them. For your agency, that means someone who can understand the client’s process, build the integration and stay accountable for how it behaves in use.

Technical fit

My strongest areas are API and webhook integration, JavaScript/Python, application state, databases and troubleshooting across the frontend and backend. I work with visual automation tools, including Make and n8n, and use code where the process needs more control. My experience includes Airtable, Supabase and HubSpot, with particular attention to validation, duplicate handling and recovery when an external service fails.

On voice, my delivered experience is with custom code and voice/telephony APIs: interpreting spoken or selected input, preserving the caller’s state and continuing through the appropriate route. My voice implementations have used custom APIs rather than the specific Retell/Vapi/Twilio stack. I can take responsibility for connecting your chosen voice platform to the operational workflow.

Work I have personally built

  • A voice AI system delivered for business use. I developed an internal application connected to a call operation, including interactive routing and state handling, then documented and transferred it to the client’s technical team for ongoing maintenance. I handled the implementation and handover myself.
  • Sofía, my own application. I have built across its React frontend, Node backend, database and AI integrations. My responsibilities include authentication, persistent state and recovering interrupted processes so the interface and backend remain consistent.
  • A local AI call assistant. I implemented audio capture, Whisper transcription, conversation context and suggested responses using a local model alongside a structured call tree. It is a prototype; the underlying branching call interface is available to try and includes session recovery and validation.

The voice and application work is presented in my portfolio. You can also try the call-tree interface.

What I would own in your receptionist example

The delivery would include the voice interaction, booking and CRM integrations, notifications, failure handling, testing and documentation. I would also take responsibility for agreed monitoring and maintenance after launch.

A booking should only be confirmed to the caller once the calendar has accepted it. If the CRM update then fails, that needs to remain visible and recoverable without creating another booking. I would include those situations, along with unavailable slots and human transfer, in the acceptance checks. That is the standard I would use before presenting the system to your client as ready.

Working arrangement

I’m interested in an ongoing contractor relationship, with an agreed scope and compensation for each client build and maintenance defined separately. I understand that you’re building the pipeline and that monthly volume is not guaranteed.

For compensation, I would quote a fixed fee against the first client brief. The CRM, telephony provider and booking rules determine the effort, so I would settle the amount with you before any work starts. The same applies to a production timeline: I would separate implementation, integration testing and the client’s acceptance checks, then give you a delivery estimate against those dependencies.

I’m available to discuss the first brief now, based in Spain (Europe/Madrid, currently UTC+2), and can work with you in English, Spanish or French. We can agree the weekly allocation around the projects you have coming in.

I’d be glad to discuss the collaboration and see how I could contribute to your team. Happy to continue the conversation here.

Zay

16

Hi Ahmed,

Answering your list in order, including the part where my answer is no.

Introduction. I am Cristian, based in Curitiba, Brazil. I am not an agency and I am not reselling someone else’s work. I run the automation side of a marketing and mentorship operation, which means I am the person who gets called when something stops at 6am, and that is where most of what I know comes from.

Strongest technical skills. n8n self-hosted in production, REST APIs and webhooks, WhatsApp (official Cloud API and unofficial), CRM integration, Postgres, LLM APIs (OpenAI and Claude) used for classification and message writing, Docker on a VPS behind a reverse proxy, and the boring layer that keeps it alive: retries, dead-letter handling, alerting and kill switches.

n8n experience. Over 100 active workflows on my own self-hosted instance, running every day against real money and real customers. Not templates, not demos.

Retell / Vapi / Twilio. No. I have never built a voice agent, and I would rather say that now than discover it on your first client. What transfers is everything after the caller speaks: intent handling, qualification logic, calendar booking, CRM writing, notification and human handoff. What I would have to learn is the telephony layer itself, and I would expect to be judged on how fast I get there, not on claiming it today.

Three things I personally built.

  1. Lead pipeline from paid ads to booked call. Meta Ads lead lands with its UTM intact, first WhatsApp touch goes out in under two minutes with no human involved, a 7 touch follow-up cadence runs, and the lead is routed by revenue band. I built the trigger, the CRM writes, the cadence and the guards. The guard that matters: the cadence halts the instant the person replies, checked twice, once before sending and once at the stage level, because checking only once is how people get messaged after they already answered.

  2. Reactivation robot over a cold base. 7 touches, per-day ceiling, business-hours window, holiday gate, and a Telegram alert to me when the queue backs up instead of failing silently. The lesson that shaped it: a failed send must not advance the cadence counter, or the contact silently loses a touch and nobody ever finds out.

  3. Shared inbox for clinics. One self-hosted panel serving multiple businesses, WhatsApp conversations with per-user login and ownership per conversation, connection monitoring that also checks the webhook is registered and not only that the socket is up. Installed for six businesses. Marginal cost measured at zero per month.

What I personally handled in all three: architecture, build, deployment, and being on call afterwards.

Demos. No public GitHub or Loom for these, because they run on client data. I can screen share a live instance and walk the actual executions, including the failures and what the guards did about them, which is more useful than a recorded happy path.

On production thinking, since you named it. The three failures I design for first are: the API that returns 200 and does nothing, the retry that duplicates a message to a real person, and the automation that keeps running after the human already took over. All three look perfectly healthy in the execution list. That is why I log the decision and not only the request, and why every flow I ship has a way to be switched off per client rather than all at once.

Rate. 45 USD per hour, or fixed price per project once scope is clear. I prefer fixed price for a first delivery, so you can judge the result rather than the clock.

Working model. Remote, async by default, with a weekly call. Comfortable in a shared repo and with someone reviewing my work.

Availability. 20 to 25 hours per week, stable and ongoing, not squeezed between other contracts.

Timezone. UTC-3. Overlaps a full working day with Europe and the whole morning with North America.

Time to build a production-ready AI receptionist like your example. For the first one, 3 to 4 weeks. Roughly one week for the voice layer, intent and data capture, one week for qualification, calendar and CRM, one week for the human handoff, retries, logging and monitoring, then a week of real calls before I would call it production. The second client of the same shape is about a week, because by then the pattern is built. Anyone quoting you a few days for the first one is quoting the happy path and leaving the edge cases for your client to find.

Happy to do a small paid pilot first so you can see how I work before committing to anything long-term.

Cristian

17

Short introduction

I build production automation: n8n workflows, LLM and voice integration, and the APIs and databases underneath them. Computer Science graduate, based in Dhaka. Most of my work has been lead pipelines, outreach systems and an AI receptionist, where the automation runs daily and someone notices when it breaks.

Strongest technical skills

n8n (self-hosted), REST APIs and webhooks, LLM integration (OpenAI, Claude, Gemini), Retell voice agents, PostgreSQL and schema design, Python and JavaScript, error handling and failure paths, Docker and Linux, Git.

n8n experience

Around a year, self-hosted on my own VPS. Production work rather than templates: scheduled triggers, conditional routing, retries, error branches, and workflows I had to keep running rather than hand over and forget.

The one I’d point at is a daily pipeline that scraped listings into Postgres, ran a cost-ordered filter chain, scored images through a vision model behind six gates, then pushed qualified records into a CRM with deduplication by phone plus name verification — because contacts at the same company often share one office line, and naive phone dedup silently merges different people into one record.

When it outgrew n8n’s 300-second execution ceiling I moved it to a queue table and partitioned the push across two scheduled runs so they couldn’t double-process.

Retell / Vapi / Twilio

Retell: I built an AI inbound receptionist on it — answers the call, works out what the caller wants, collects name, phone and service, qualifies the lead, and books the appointment, with the result written back to the CRM. It sat on top of a contractor lead pipeline I’d built in n8n, so the voice agent and the data layer behind it were one system rather than two things bolted together.

Twilio: SMS outreach on a registered 10DLC number, with compliance filtering running before send rather than after, and LLM reply classification on inbound.

Vapi: Haven’t used it. Same category as Retell, so I’d expect to pick it up quickly, but I’m not going to claim it.

2–3 projects I personally built, and what I personally handled

1. AI inbound receptionist + contractor lead pipeline. Retell voice agent on top of an n8n pipeline that scrapes state contractor-license registries, enriches through Apollo, and tier-scores on license class, reviews, headcount and geography to surface the top ~30% for follow-up. I handled the pipeline, the scoring model, the CRM integration and the voice agent’s booking path.

2. Daily property-lead pipeline. Built during an AI operations internship, so the system belongs to my employer — I did the building rather than own it. Mine specifically: the Postgres schema including dedup hashes, indexes and retention, the filter chain and its cost ordering, the vision-scoring stage, the CRM push and dedup logic, and the queue-partition fix for the execution ceiling. Measured funnel on one production day: 729 records in, 25 qualified out. Ordering the free database filters ahead of every paid API call cut spend and external request volume by roughly 80% and held running cost near $105/month.

3. wallop. Mine entirely. A load-testing CLI published on PyPI — asyncio worker pool, live p50/p95/p99 from a rolling window, ~40 tests green across a 3-OS by 2-Python CI matrix. GitHub - siam-hossain9/wallop: Terminal HTTP load generator that draws every request as a physics particle in a half-block ANSI framebuffer while reporting live RPS and p50/p95/p99; asyncio and aiohttp, one dependency · GitHub

Links

siam-hossain9 (Siam Hossain) · GitHub · siamsportfolio.vercel.app · I have a portfolio PDF with n8n workflow screenshots and architecture diagrams, happy to send it over.

Expected compensation / rate

$30/hour for support, debugging and maintenance. For defined builds I’d rather quote per project so you know the number before work starts.

Preferred working model

Project-based for defined builds, hourly for debugging and maintenance. I’m comfortable with no guaranteed monthly volume, as long as each engagement is scoped and agreed before it begins — which is what you’ve described.

Availability

Available now.

Timezone

UTC+6 (Dhaka). Comfortable with European hours and US Eastern mornings.

How long for a production-ready AI receptionist like your example

A working demo: 3–5 days. Production-ready: 2–3 weeks.

The gap between those two is the entire job, and it’s the list you already wrote. Calendar booking that can’t double-book when two calls land in the same minute. What happens when the CRM write fails after the caller has already been told the appointment is confirmed. Callers who interrupt, go silent, or give a name the transcription mangles. The human-transfer path tested for when the human doesn’t pick up. Retries that don’t create a second booking. Logging good enough to reconstruct a call that went wrong three days ago.

A demo answers the phone. Production survives the calls the demo never got.

18

Hi Ahmed

This role immediately stood out to me because I already build the kind of end-to-end automation systems you described, particularly around n8n, APIs, AI agents and voice automation.

1. Introduction

I’m Adeleke Adenike Esther, an AI Automation Specialist based in Nigeria (WAT, UTC+1). I build and troubleshoot AI-powered workflows for business processes and enjoy taking systems from the initial logic through testing and deployment.

2. Strongest technical skills

n8n, REST APIs, webhooks, JavaScript, JSON/data transformation, OpenAI/Claude/Gemini APIs, AI agents, Vapi, Airtable, Google Sheets, Supabase/PostgreSQL, API testing, error handling, retries and workflow monitoring.

3. n8n experience

n8n is one of my primary automation tools. I use it to orchestrate multi-step workflows, connect APIs without native integrations, process and transform data, route AI decisions, trigger notifications and build error workflows. I also use http inside n8n when the built-in nodes are not enough.

4. Vapi / Retell / Twilio

I have hands-on experience with Vapi and have built a logistics voice automation system around it. I have also worked with Twilio. I have not used Retell extensively, but I’m comfortable learning new API-based platforms quickly.

5–7. Projects I personally built

Logistics Voice Automation: Vapi + n8n + REST APIs + JavaScript + Airtable. I personally handled the workflow architecture, voice-agent flow, API integrations, data validation/normalization, JSON transformation, notifications, retry logic and error handling.

Immigration Consultancy AI Customer Support & Intake System: I built the website and complete n8n backend for an AI-powered customer support system that can handle common enquiries, provide instant responses, collect and structure client information, and intelligently escalate cases when human assistance is required. The workflow also handles team notifications and client confirmations, giving the consultancy an end-to-end automated support and intake process.

HR Recruitment Dashboard & Document Handling Workflow: I built an HR recruitment system that helps organize candidate information and handle recruitment documents through a structured workflow. The system centralizes candidate records, processes submitted documents such as CVs/resumes, and connects the information to the recruitment workflow, reducing the manual work involved in reviewing, organizing and moving candidate information through the process.

Portfolio: https://adelekeestherportfolio.lovable.app/

8. Expected compensation

I’m open to discussing the rate based on the scope and level of responsibility. For ongoing technical work, I would be comfortable around $15–$20/hour, with fixed project pricing also possible once the scope is clear.

9. Preferred working model

Remote, long-term collaboration. I’m comfortable working independently while communicating closely with the team.

10. Availability

Available to start immediately and able to commit consistently as client work comes in.

11. Timezone

WAT (UTC+1).

12. Production-ready AI receptionist timeline

For a system similar to your example, I would estimate 1–2 weeks for a solid production-ready v1, assuming access to the required CRM, calendar, telephony and business requirements. I would first map the call flow, integrations and failure paths, then build, test edge cases, deploy and monitor it.

What particularly interests me about this opportunity is the long-term aspect. I’m not looking to simply complete an automation and move on. I want to contribute to the technical side, solve real client problems and help build reliable systems as the agency grows.

Thank you for considering my application, looking forward to hearing from you.

19

Hi, I’m Haider, a Computer Engineering student and Software Developer with hands-on experience building AI agents, automation workflows, backend systems, APIs, and database integrations. My strongest technical skills are Python, Node.js, C#, React, SQL Server, REST APIs, webhooks, Git, and OpenAI/LLM integrations.

I have personally built an AI inventory tracking agent that handles inventory-related data and automates the tracking process, as well as an AI call-handling agent that can communicate with callers and handle their requests through an automated workflow. These projects gave me practical experience with AI agents, LLM logic, data handling, automation, and building workflows around real business use cases.

I also have professional software development experience from my internship at OMA Emirates, where I worked on a full-stack system using React, Node.js/C#, SQL Server, REST APIs, CRUD operations, and backend/database integration. I personally handled the frontend and backend integration, API communication, database operations, and debugging.

I’m familiar with working with APIs, webhooks, databases, authentication, error handling, and troubleshooting. I haven’t worked extensively with every platform listed in the post, particularly Retell/Vapi and Twilio, but I already have experience building AI agents and connecting different systems, so I’m confident I can quickly adapt to those platforms.

20

Hi Ahmed,

I’m interested in the technical side of what you’re building, particularly the n8n, API/webhook, CRM, data handling, validation, and reliability work.

My current hands-on experience comes from independent projects rather than client production work, so I want to be transparent about that.

My strongest areas are:

  • n8n
  • REST APIs, HTTP, JSON, and webhooks
  • Gmail and Google Sheets automation
  • AI classification and structured extraction
  • Airtable and Slack automation
  • HubSpot API/CRM integration
  • validation, duplicate prevention, logging, error handling, and human-review paths

A few projects I personally built and tested:

  1. An AI-powered Gmail workflow in n8n that classifies incoming emails, extracts structured information, checks approved data in Google Sheets, routes uncertain cases to human review, and creates draft replies with duplicate prevention and logging.

  2. A HubSpot developer-test integration covering API-based CRM create/update operations, validation, and error handling.

  3. An Airtable + Slack lead automation with validation, routing, notifications, logging, and repeat-input handling.

I handled the workflow design, implementation, testing, debugging, and reliability logic myself.

I have not yet worked with Retell, Vapi, or Twilio in a client production environment, so I would not overstate that experience.

For that reason, I’d prefer to start with one small, clearly scoped paid implementation ticket or trial on the n8n/API/CRM/reliability side before you commit to anything larger.

For example, a $75–100 fixed trial could cover one bounded workflow branch, API mapping, debugging task, validation/error-handling improvement, or similar technical ticket.

If that goes well, I’d be interested in continuing as the agency and workload grow.

Working model: remote, project/contract initially, with the option to develop into an ongoing collaboration.

Availability: available for a small initial paid task now.

Timezone: Europe/Berlin.

For a complete production-ready AI receptionist, I would want to inspect the exact voice provider, telephony setup, CRM/calendar requirements, handoff rules, failure paths, and monitoring requirements before giving you a reliable estimate. Since Retell/Vapi/Twilio are not yet part of my proven production experience, I would rather demonstrate my current strengths first than give you an aggressive estimate I cannot defend.

I can also share screenshots or a concise walkthrough of the projects above if useful.

Daniel