Open to opportunities

ken doki

@kendoki

Message

Senior Applied AI Engineer building cost-aware, citation-grounded LLM systems with evaluation-first delivery.

Malaysia

Message

What I'm looking for

I’m looking for an Applied AI role where I can build grounded, citation-rich LLM systems with rigorous evals, cost-aware routing, guardrails, and production-grade rollout. I value engineering partnership, measurable quality gains, and safe iteration.

I’m a Senior Applied AI Engineer who builds grounded, production-ready LLM systems with rigorous evaluation and safety guardrails. At Etiqa Insurance, I delivered an internal claims operations assistant that cut routine policy lookups by ~90%, moving from minutes of cross-referencing to a single grounded query with citations.

I designed the retrieval pipeline end to end on Azure OpenAI, including section-aware chunking, hybrid search (pgvector + Postgres full-text), reciprocal rank fusion, and cross-encoder reranking—improving precision to ~86%. I also built a supervisor/worker agent layer with typed tools and persisted per-request state, and I exposed structured lookup tools via an MCP server for reuse.

I care deeply about reliable delivery: I implemented two-stage routing to keep ~70% of traffic on a cheaper model without quality regression, and I shipped versioned prompts with A/B and canary rollout plus auto-rollback. I paired LLM-as-judge scoring with CI regression gates, enforced an augmentation-only boundary to prevent unsafe outputs, and added multi-tenant isolation with Postgres Row-Level Security alongside OpenTelemetry tracing and cost dashboards.

Experience

Work history, roles, and key accomplishments

Senior Applied AI Engineer

Etiqa Insurance

Apr 2024 - Nov 2025 (1 year 7 months)

Built an internal claims operations assistant that reduced routine policy lookups ~90% by consolidating cross-system searches into a single grounded, citation-backed query run daily in production with human review. Designed an Azure OpenAI retrieval and agent pipeline using hybrid pgvector + Postgres full-text search, cost-aware model routing, MCP tool exposure, and eval-gated prompt rollouts with

RAG Agent orchestration Azure OpenAI Hybrid Search (Pgvector + Postgres Full Text)Reciprocal Rank Fusion MCP Servers Structured Outputs (Pydantic)OpenTelemetry

Senior Applied AI Engineer

Freelance

Apr 2023 - Mar 2024 (11 months)

Built and shipped an AI chatbot platform that answers questions over company documents, taking prototypes to production across a Python RAG pipeline and a TypeScript product layer. Implemented streaming (SSE) chat with inline citations, reduced per-query cost ~40% via multi-model routing and semantic caching, and shipped Stripe billing plus integrations (Slack, Intercom, Notion, Drive) with an adm

RAG Document Ingestion & Chunking Hybrid Retrieval (Rank Fusion + Reranking)Streaming (SSE)Model Routing Token Level Cost Logging Stripe Billing

Senior Software Engineer

Accenture

Nov 2020 - Mar 2023 (2 years 4 months)

Took a document intelligence/logistics product to production with a two-service architecture combining a TypeScript product layer and a Python OCR/NLP service, connected via typed contracts to prevent schema drift at build time. Rebuilt operator review workflows for higher throughput and led a composable travel and expense workflow engine with template-driven state-machine transitions gated on for

TypeScript Python (OCR NLP)CI Regression Prevention (Schema Drift Control)Form Validation

Software Engineer

Sunway

Nov 2019 - Oct 2020 (11 months)

Built the backend API, real-time ops dashboard, and live attendance monitoring (Socket.IO with QR and GPS-verified check-in) for events security operations. Developed a constraint-based scheduling engine using SQL and PostGIS pre-filtering with weighted scoring, cutting scheduling time from hours to minutes and preventing double-bookings at the database level.

SQL PostGIS Constraint Based Scheduling QR Based Check In Back end API development Dashboards