我想尋找一位專業人士

Jobs
1

我正在尋找懂印尼語的 n8n 工程師／AI 自動化工程師，用於審計和修復生產工作流程。

背景：

  • 系統運行在 n8n 上

  • 整合：Telegram、WhatsApp/Messenger flow、AI agent、Google Drive/Pinterest、webhook、報告

  • 問題：工作流程還不穩定，需要端到端審計、錯誤處理、日誌記錄、備用方案和修復而不破壞現有系統

  • 不是從零開始創建示例工作流程的工作

  • 必須能夠讀取複雜工作流程、追蹤執行、修復漏洞、建立測試用例，並在聲稱完成之前提供證據

要求：

  1. 具有 n8n 生產工作流程的實戰經驗

  2. 了解 webhook、API、憑證安全、錯誤處理、重試、日誌記錄

  3. 了解 AI agent／LLM 工作流程

  4. 會印尼語

  5. 願意透過 staging/sandbox 工作，而不是直接修改生產環境

  6. 能夠提供變更文檔和測試證據

請發送 n8n 作品集、您曾經處理過的案例示例，以及您進行審計的估計方式。

2 個讚
4

Halo ady_sihombing,

Saya tertarik dengan project audit n8n production workflow Anda. Latar belakang saya:

Pengalaman:

  • n8n production workflow dengan integrasi Telegram, WhatsApp, API, webhook, AI agent/LLM
  • Error handling, retry logic, logging, credential security
  • Audit sistem kompleks yang sudah berjalan tanpa mengganggu production

Metodologi audit yang saya tawarkan:

  1. Duplikasi workflow ke staging/sandbox
  2. Tracing execution path end-to-end
  3. Identifikasi bottleneck, gap error handling, dan single point of failure
  4. Implementasi: retry with exponential backoff, fallback mechanism, structured logging, alert notification
  5. Test case untuk setiap node kritis
  6. Dokumentasi lengkap + bukti test sebelum deploy ke production

Saya menawarkan fixed price $750 untuk full audit + fix. Estimasi 3-7 hari tergantung kompleksitas workflow.

Jika tertarik, silakan kirim akses ke staging/sandbox. Saya bisa mulai analisis dalam 1x24 jam.

Terima kasih.

5

Olá, Ady! Tudo bem?

Tenho experiência com workflows n8n em produção, incluindo integrações com WhatsApp, AI agents/LLMs, Google Drive e webhooks, exatamente o tipo de stack que você descreveu. Falo português nativamente e já trabalhei em projetos que exigiam correção de bugs sem quebrar sistemas já em funcionamento.

Portfólio: https://www.jeffersoncampina.tech/

Exemplos de casos relevantes:

  • Desenvolvi a Lia, assistente de recrutamento com IA integrada a WhatsApp (Chatwoot), OpenRouter API e Supabase. Tive que diagnosticar e corrigir um bug de produção onde o agente “alucinava” confirmações de ações que não tinha executado resolvi movendo lógica determinística para nós nativos do n8n e adicionando validação pós-execução.
  • Construí pipelines de triagem com AI Agent estruturado (output em JSON), incluindo debugging de nós com respostas de API em formatos inesperados (markdown envolvendo JSON, mismatches de tipo de nó).
  • Tenho experiência com arquitetura multi-agente mais robusta (LangGraph, FastAPI) o que me ajuda a entender fluxos complexos de AI agent mesmo quando construídos só em low-code.

Como eu faria a auditoria:

  1. Mapeamento: leitura completa do workflow, documentando cada nó, trigger, credencial e dependência externa (sem alterar nada em produção).
  2. Tracing de execuções: analisar logs/execuções passadas (sucesso e falha) para identificar onde o fluxo quebra ou se comporta de forma inconsistente.
  3. Ambiente de staging: replicar o workflow em sandbox com credenciais de teste antes de qualquer mudança.
  4. Correções incrementais: error handling, retry logic, logging e fallback, testados isoladamente por trecho do fluxo.
  5. Test cases: criar casos de teste cobrindo os cenários críticos (falha de API, timeout, resposta inesperada do LLM).
  6. Documentação e comprovação: relatório com o que foi alterado, evidência de teste (prints/logs de execução) antes de considerar qualquer etapa concluída.

Fico à disposição para conversar mais sobre o escopo e prazo.

6

Upfront: I don’t speak Indonesian, so if that’s non-negotiable, ignore this and good luck with the search.

If English works, this is the kind of work I do. How I’d approach an audit like yours: read the workflows and map every external call and its failure mode first, then reproduce the failures in a staging copy rather than touching production. Fixes go in one at a time, each proven with a before and after execution log, so “fixed” is something you can see instead of something I claim.

Deliverables would be the fixed workflows, a dedicated error workflow that alerts with enough context to know what broke without opening n8n, retries with backoff on the external calls, and a written note per change.

Happy to send examples of similar work. Either way, the evidence-before-claiming-completion standard you wrote is the right one, and I’d hold to it.

7

Hi Ady, I realize you’re looking for someone who speaks Indonesian. While I don’t speak it fluently, I do have experience auditing and improving production n8n workflows involving AI agents, APIs, webhooks, error handling and workflow reliability. I’ll send you a DM in case my technical background is helpful.

8

Hello Ady. I’m Misha, an n8n and AI automation developer. I understand your production workflows are unstable and need a careful audit without risking the running system. I would start by replicating your environment in a staging sandbox to trace each workflow—Telegram, WhatsApp, AI agent calls, Google Drive, and webhooks—identifying where error handling, fallbacks, and logging are missing or failing. I would fix each failure point systematically, add retry logic and credential security checks, then run test cases with documented evidence before marking anything complete. My approach keeps your production untouched until every fix is proven solid. I speak Indonesian and can communicate the changes clearly. Portfolio: https://mikedevai.netlify.app/ | Telegram: https://t.me/hely_chatbots | WhatsApp: +375293761570

9

Let’s connect