Senior Full Stack Developer

Key Responsibilities

• Design, build and ship features end-to-end across backend (NestJS, GraphQL, PostgreSQL) and frontend (Next.js, React)
• Translate business requirements in HR, accounting, sales and operations into correct, well-modelled domain logic — and question the requirement when it's incomplete
• Own features from scoping to production: surface open decisions early, estimate honestly, sequence the work, and deliver the unglamorous parts too
• Drive AI coding agents effectively: write clear specs, keep them on-scope, verify their output rigorously, and improve the conventions that make them more reliable
• Design and evolve the database schema for a multi-tenant system with Postgres row-level security; write safe, reversible migrations
• Build and maintain integrations (WhatsApp/Meta, Google Workspace, accounting vendors, payment gateways)
• Review pull requests with a sceptical eye; contribute to our automated guardrails (convention checks, RLS smoke tests, CI)
• Communicate progress, risks and trade-offs clearly to the founder and non-technical stakeholders

Requirements

• Degree in Computer Science / Software Engineering or equivalent practical experience
• Minimum 5 years of professional software development; at least 3 years with TypeScript across both backend (Node.js) and frontend (React)
• Strong hands-on experience with relational databases (PostgreSQL preferred): schema design, migrations, transactions, query performance
• Solid experience with NestJS or a comparable Node framework, and with Next.js/React
• Proven business-logic ability: you have built systems where correctness had real consequences — accounting periods, approval workflows, leave/payroll rules, invoicing, inventory
• Project handling: able to take a loosely defined feature from a founder to a scoped plan and a shipped result, and to manage your own priorities without a PM
• Serious AI-assisted development experience: daily use of Claude Code, Codex, Cursor or similar as a primary workflow, with the judgement to know when its output is wrong
• Comfortable in a monorepo with CI/CD and automated checks; you treat a green build you don't understand as unverified
• Must be able to communicate in both English and Chinese to serve Chinese-speaking customers.

Added advantage

● Multi-tenant SaaS with database-enforced tenant isolation (RLS), OpenFGA or similar authorization

● GraphQL at scale (code-first schema, codegen, N+1 handling)

● Building with LLMs in production: RAG, agents/tool use, evaluation, cost control (we use LiteLLM, pgvector, multiple providers)

● Malaysian business context: MyInvois e-invoicing, local HR/leave/payroll rules, WhatsApp-first customer communication

● Python for data/ML services; React Native / Expo; Redis/BullMQ