Clarify
Write the spec. Name the risk. Set the success check.
Haoming Koo · Applied AI Engineer
AI systems that survive reality.
I build production LLM systems for messy work: RAG, agent workflows, evals, observability, and product UX. Based in Singapore.
Open to full-time AI engineering roles and focused client projects. Singapore-based; open to remote teams globally.
Recruiter search: Haoming Koo, AI Engineer Singapore, Production LLM Systems, LangGraph, RAG, Evals, LLM Observability.
$600M+business impact
8live products
3,000+engineers aligned
4global fabs
$600M+ is the aggregate value of programs I led, co-led, or contributed to as tracked in Micron internal program reviews.
Role fit
Same builder. Different lens.
Frontier team, startup, or client project. The value is still the same: clarify, build, verify, ship.
Built for model-facing product work.
I can move between model behavior, data plumbing, evals, observability, UI, and user feedback.
Operating loop
From fuzzy ask to working system.
This is the pattern I bring to AI work. It keeps demos close to production.
Ground
Use retrieval, tools, data, and source traces.
Orchestrate
Build LangGraph-style flows with human review.
Verify
Add evals, observability, and failure handling.
Proof, not adjectives
Useful in rooms where trust matters.
I have worked around fabs, engineers, models, dashboards, and public users. The common thread is reliability.
Enterprise work with measurable impact.
Cross-site transformation across Singapore, Boise, Hiroshima, and Taichung.
RAG and agent workflows that can be reviewed.
Specs, retrieval, validation gates, traces, evals, and handoff.
Products people can open and inspect.
Job search, markets, travel, benefits, weather, compliance, and data tools.
Testimonials
What former leaders noticed.
The pattern is the same across roles: clear communication, structured problem solving, and calm delivery.
Excellent communication skills and an aptitude for building consensus among cross-functional and cross-cultural teams. A trusted partner and someone I can rely on in challenging situations.
Distinguished Member of Technical Staff, Micron Technology
Approaches problems with structure, curiosity, and a commitment to long-term improvement. Energy, analytical rigour, and a collaborative spirit in complex environments.
Senior Manager, Micron Technology
Selected work
Live products with real constraints.
These are the fastest proof points. They show range without asking for trust upfront.
Top project · LLM product
Job Hunter SG
Singapore job search product with RAG matching, ATS scoring, role-fit keyword checks, and fact-checked resume edits.
Interactive signal
Hover a product card to inspect the system signal.
Market workflow
Trader Koo
Review workflow for market signals, reports, paper-trade records, and source freshness. Not financial advice.
MCP server
Japan in Seasons
Travel intelligence server with 12 tools, 1,700+ places, seasonal signals, and a 100/100 Smithery score.
Source-backed UX
Amex Explorer
Benefit explorer with maps, official-source tracking, stale-data alerts, and manual review gates.
Vision product
Photo Compliance Studio
Passport-style photo checks with face landmarks, quality diagnostics, and guided correction flows.
Private work, public-safe
Complex planning, described cleanly.
Some current work cannot be shown in detail. The capability can still be explained.
Enterprise planning workflow
From request to reviewed plan.
Specs, retrieval, tool orchestration, validators, human review, traces, and auditable recommendations.
Observability
Measure the agent, not just the answer.
Langfuse and Arize Phoenix style tracing, eval datasets, scorecards, and failure review.
Spec-driven development
Make ambiguity testable.
Clear requirements, acceptance checks, edge cases, and a product surface users understand.
Story
Lab bench to AI products.
The path looks unusual. That is the advantage.
Micron Technology
Cross-site digital transformation. Four fabs. USD 600M+ business impact.
Production ML
Vision, sequence modelling, RAG, inference, CI/CD, APIs, and UIs.
Independent shipping
Jobs, markets, travel, benefits, weather, compliance, EDA, and wine data.
High-agency AI engineering
Model-facing, user-facing, and operationally real.
Client work
Small scope, real output.
I am open to focused paid projects. The goal is a working system.
AI workflow
LLM tools for teams
RAG, agent orchestration, evals, prompt systems, observability, and guardrails.
Data product
Dashboards that decide
Data ingestion, analytics, forecasting, anomaly detection, and operational UX.
Automation
Monitoring systems
Source tracking, scheduled jobs, data quality checks, alerts, and review queues.
Next conversation
Send the problem. I will bring the system.
Hiring for applied AI, ML engineering, AI deployment, AI product engineering, or focused client work? I can map the first version quickly.