Work Projects

Professional and research collaborations where I work on reliable, secure, and deployable AI systems.

RESISTANT

Zero-trust avionics research project. I contribute as tech lead and core developer, designing attack-graph modeling with partially observable MDPs, particle-filter risk assessment, and tabular foundation models for adaptive security decisions.

Security Risk modeling POMDPs TabPFN

SPARTA

Academic collaboration on social-media monitoring and analysis. I serve as project lead, with work on political stance detection and malicious-account detection using vision-language embeddings and graph neural networks.

Project lead Social media analysis VLM embeddings GNNs

Featured Work

Selected projects that connect applied ML with trust, evaluation, and practical deployment.

NaturalQL - LLM API NL-to-SQL
NaturalQL - LLM API NL-to-SQL

LLM API-powered natural-language to SQL demo with schema-bounded prompting, static validation, and safe execution.

LLM APIs OpenAI API DuckDB Streamlit sqlglot Python

Vision-Only Chess Tracker
Vision-Only Chess Tracker

Chess move tracking from raw camera pixels on ordinary boards, across arbitrary board poses, without instrumented hardware.

Computer Vision PyTorch Blender OpenCV Python

Experiments / Playground

Older and smaller projects I keep around as sketches, exercises, and technical notes.