Mid-Level Full Stack Developer (AI Customer Service Platform)
Salary: RM 6,000 – RM 7,000 per month
Key Responsibilities
Build, maintain and debug n8n workflows, including sub-workflows, Execute Workflow nodes, AI Agent nodes with tool calling and queue mode.
Trace execution logs to the root cause of a problem instead of patching symptoms.
Design and tune LLM prompts for structured JSON output, intent classification and confidence thresholds, and move logic out of prompts and into code where that is more reliable.
Manage per-merchant configuration, credentials, Redis caching and LLM provider routing and fallback across tenants.
Maintain integrations with Chatwoot (inboxes, agent bots, HMAC identity validation), Telegram, web livechat and merchant APIs and webhooks.
Work on receipt parsing, bank account matching and reconciliation for deposit and withdrawal flows.
Build and maintain logging, reporting and monitoring so failures are visible quickly.
Make sure the system fails to a human agent and never to a wrong answer.
Write and maintain runbooks, workflow documentation and architecture notes.
Make careful, reversible production changes: test before release and have a rollback plan.
Requirements
3–5 years of full stack or backend development experience.
Hands-on n8n experience: Code nodes, Execute Workflow, expressions, paired items and null handling.
Strong JavaScript skills (data transformation, string matching, debugging).
PostgreSQL: joins, jsonb, upserts (ON CONFLICT) and safe UPDATE/DELETE practice.
REST API integration: authentication schemes, headers, webhooks and error handling.
Experience with an LLM API (OpenAI, OpenRouter or similar), including prompt engineering for structured output.
Comfortable with Git and systematic debugging.
Strong problem-solving skills and ability to work independently.
Fluent in English and Bahasa Malaysia.
Nice-to-Haves
RAG fundamentals: embeddings and vector similarity.
Redis caching (TTLs, invalidation).
Chatwoot administration.
AWS basics (ECS, NAT egress IPs, IP whitelisting).
Experience with multi-tenant system design.
Mandarin