Full-Stack AI Engineer

Full-Stack AI Engineer

About the Role

You will be one of the first core builders on the AI team — owning end-to-end delivery of AI products from prototype to production. This role spans the full AI engineering stack: LLM applications, RAG, multi-agent systems, tool calling, backend services, workflow automation, evaluation, deployment, and production observability.

This role is for someone who has shipped real AI products to real users, thrives in ambiguity, and understands that excellent AI requires strong engineering discipline, not just good prompts — including evals, guardrails, UX judgment, cost control, and measurable outcomes. You will help establish the foundational engineering culture for AI at the company.

What You Will Do

Build & Architect

●      Design and build end-to-end AI products: LLM applications, RAG pipelines, multi-agent systems, tool-calling workflows, and backend services integrated with internal APIs, databases, and enterprise tooling.

●      Develop agentic workflows for business use cases including customer service, fraud & risk operations, knowledge search, decision support, and productivity automation.

●      Implement retrieval systems with chunking strategies, embedding pipelines, hybrid search, reranking, grounding, and citation logic; build prompt pipelines with fallback flows, input/output guardrails, and human-in-the-loop escalation.

●      Prototype quickly, validate assumptions with real users, and harden successful pilots into reliable, production-grade systems.

●      Establish early AI engineering standards covering observability, auditability, privacy, security, cost governance, and safe rollout patterns (canary, shadow mode, feature flags).

Evaluate & Operate

●      Build evaluation harnesses using golden datasets, regression tests, LLM-as-judge scorers, human preference feedback, and production telemetry to measure quality before and after deployment.

●      Monitor model behavior across latency, cost, hallucination rate, tool failure, retrieval quality, context window usage, and user satisfaction signals; own incident response and root-cause analysis for AI failures.

●      Instrument AI systems with structured traces, spans, and metadata to support debugging, compliance auditing, and continuous improvement.

●      Drive iterative quality improvement loops: identify failure modes, A/B test interventions, and track measurable outcomes against business KPIs.

●      Collaborate cross-functionally with product, business, data, risk, compliance, and engineering teams to translate ambiguous problems into working AI systems with clear success criteria.

What We Are Looking For

Must Have

●      Strong software engineering background in backend or full-stack development; experience building production APIs, microservices, databases, queues, and deployment pipelines.

●      Hands-on experience building LLM applications using provider APIs (OpenAI, Anthropic, Gemini, Azure OpenAI, or similar), including prompt engineering, context management, structured outputs, and function/tool calling.

●      Experience with RAG systems: document ingestion, chunking, embedding, vector databases (e.g., pgvector, Qdrant, Weaviate, Pinecone), retrieval quality tuning, reranking, and grounding.

●      Practical command of AI system failure modes — hallucination, retrieval drift, tool failure, context overflow, latency degradation, cost blowout, and unsafe outputs — and how to design against them.

●      Strong Python skills; TypeScript/JavaScript a strong plus. Ownership mindset with the ability to operate with minimal structure and communicate clearly across technical and non-technical audiences.

Strong Plus

●      Experience in fintech, banking, payments, fraud, risk, compliance, or other regulated industries with data governance and auditability requirements.

●      Experience with agentic frameworks (LangGraph, LlamaIndex, AutoGen, CrewAI, DSPy, Semantic Kernel) and cloud infrastructure (AWS, Azure, GCP, Kubernetes, managed vector/ML services).

●      Experience with LLMOps tooling: evaluation frameworks (RAGAS, DeepEval, Promptfoo), observability platforms (LangSmith, Phoenix/Arize, Helicone), and fine-tuning workflows (SFT, DPO, LoRA/QLoRA).

●      Experience with enterprise security requirements: authentication, RBAC, audit logging, secrets management, PII detection, data residency, and PDPA/GDPR compliance.

●      Experience building multilingual AI products (Thai and English a strong plus) and optimizing LLM cost, latency, and throughput at production scale.