Open to opportunities

Christopher Lui

@christopherlui

Message

I’m an AI/ML engineer shipping production LLM pipelines, RAG systems, and agentic workflows with measurable outcomes.

United States

Message

What I'm looking for

I’m looking to build end-to-end production LLM systems—RAG, agentic pipelines, and MLOps—where I can optimize for measurable impact, reliable deployments, and continuous evaluation in collaboration with product teams.

I’m an AI/ML engineer with 8+ years building production LLM pipelines, RAG architectures, and agentic systems across healthcare and enterprise domains. I focus on shipping measurable outcomes—not just models—by designing robust orchestration, evaluation, and guardrailed deployments.

Most recently, I built a multi-agent orchestration framework with LangGraph that reduced average query resolution time by 38%, and I deployed a production RAG pipeline handling 2M embedded document chunks while cutting P95 retrieval latency from 1.9s to 0.6s through index pre-warming. I’ve also delivered document intelligence with Azure Form Recognizer (reducing manual review volume by 62%), fine-tuned LLMs with QLoRA (improving F1 by 11 points), and used model evaluation harnesses to justify inference cost savings of $120K/year.

Experience

Work history, roles, and key accomplishments

Current

Senior AI/ML Engineer

Current

Beejern

Apr 2024 - Present (2 years)

Designed a LangGraph-based multi-agent orchestration framework that reduced average query resolution time by 38% across internal healthcare data workflows. Built a production RAG pipeline (Pinecone + GPT-4o) processing 2M+ document chunks and reduced P95 retrieval latency from 1.9s to 0.6s.

LangGraph Pinecone GPT 4o QLoRA Fine Tuning Azure Form Recognizer fastAPI Docker Kubernetes

Machine Learning Engineer

Mercury Development

Sep 2022 - Mar 2024 (1 year 6 months)

Built a FAISS semantic search system and improved recall@10 from 0.71 to 0.89 by adding reranking after retrieval ablations. Developed a fine-tuned BERT NER contract clause extraction API and improved ticket routing by fixing an ingestion class-imbalance bug, raising accuracy to 91.3%.

Faiss Sentence Transformers Fine Tuned BERT NER fastAPI AWS Sagemaker Postgres Streamlit PyTorch SQL Cross Encoder Re Ranking

ML / NLP Engineer

Underguard

Jan 2020 - Aug 2022 (2 years 7 months)

Implemented a real-time fraud detection system processing 180K daily transactions and reduced false positives by fixing feature leakage, improving precision at threshold from 0.61 to 0.84. Built an NLP policy document parser with SpaCy to extract structured coverage terms from PDFs, reducing manual data entry time by 70%.

PyTorch scikit learn Flask AWS EC2 Active Learning Celery PostgreSQL

Data / ML Analyst

Akamai

Apr 2017 - Dec 2019 (2 years 8 months)

Built predictive CDN traffic models forecasting edge load 72 hours out and improved RMSE by 23% by adding content-category and referral-source signals. Automated weekly anomaly reporting across 14 data pipelines, replacing a manual 4-hour workflow and saving ~200 analyst-hours per quarter.