Looking for a job

Richard Huang

@richardhuang

AI Engineer building multi-agent LLM systems, including a LangGraph trading research pipeline with point-in-time backtesting.

United Kingdom

Message

I developed a LangGraph-based multi-agent trading research system that coordinates five specialised agents and retrieves live market data through GPT-4o and Claude Sonnet APIs. I extended it into a Python pipeline for backtesting swing-trading strategies while enforcing point-in-time data integrity.

For a freelance AI engineering project, I built a multi-agent vision pipeline that turns hand-drawn survey sketches into CAD-ready drawings. Its evaluation framework included 450+ tests, and mean room IoU improved from 0.605 to 0.799 across 20 buildings.

As an AI Engineering Research Assistant, I developed a Socratic RAG tutoring system for R for Data Science using GPT-4o-mini and vector retrieval. I containerised it with Docker and tested it with up to 100 simultaneous users, identifying an external API rate limit as the primary performance bottleneck.

Experience

Work history, roles, and key accomplishments

AI Engineering Research Assistant

Socratic RAG Tutoring System for R for Data Science

Jun 2026 - Aug 2026 (2 months)

Architected and developed a full-stack RAG tutoring system in R/Shiny, integrating GPT-4o-mini with embedding-based vector retrieval. Productionised the application as a containerised service using Docker and Shiny Server, and conducted load and concurrency testing for up to 100 simultaneous users.

R Shiny GPT 4o Mini RAG DuckDB Docker Structured logging Claude Code Testing

Junior Research Assistant

University of Edinburgh Data Science Unit

May 2024 - Jul 2024 (2 months)

Collaborated within a large multidisciplinary team to develop weakly-supervised models on annotated pre/post-flood satellite images. Reviewed causal generative literature and collaborated with senior researchers to debug causal disentanglement failures.

Machine Learning Data Engineering Causal Inference Model Evaluation