Work
Selected production engagements. Outcomes first, then the architecture behind them.
Multimodal LLM extraction that raised accuracy from 84% to 88% and cut manual review overhead by around 60%.
An agentic natural-language interface that lets underwriters query cases, policies, and guidelines in under 30 seconds.
An eight-stage production platform that turns messy scans into structured data: 84% end-to-end, later 88% with an LLM layer.
Local LLM Deployment & Serving
Self-hosted LLMs on owned GPUs serving two production pipelines, cutting cost per document from ~$0.35 to ~$0.10.
LLM Provider Benchmarking
A price-vs-accuracy evaluation harness that routes production traffic to the right model for each task.
CI/CD & Flutter Release Pipelines
Automated build, test, and deploy for multi-product stacks including Flutter Android, from manual releases to weekly cadence.
Advisify
AI immigration assistant with personalized pathway recommendations grounded in CV and preferences.
Aceprep
Structured O/A Level past-paper platform with scheme-grounded Q&A and answer validation.
