Portfolio
Projects and systems organized by the evidence pipeline: Extract → Validate → Surface.
🔍 Extract — Pulling Signal from Noise
TrackXY
Computer vision tracking system. Extracts object trajectories from noisy video feeds using custom tracking algorithms. Built from the ground up for environments where off-the-shelf trackers fail.
geo-csg
Constructive Solid Geometry from geospatial data. Extracts real-world objects from point clouds and photogrammetry, converting them to structured CSG representations for CAD and analysis.
antisci
Natural language processing pipeline for extracting structured insights from unstructured scientific and technical documents. Named entity recognition, relation extraction, and document classification.
smolpaper
Minimalist scientific paper extraction tool. Pulls structured metadata, citations, and full text from academic papers in PDF and HTML formats. Designed for researchers who need clean data fast.
Web Scrapers
A collection of resilient, production-grade web scrapers for diverse targets — e-commerce, government data portals, scientific databases, and more. Anti-detection, proxy rotation, structured output.
Reverse Engineering
Protocol and binary reverse engineering projects. Extracting data from proprietary formats, undocumented APIs, and legacy systems. Building bridges between closed systems and modern tooling.
✓ Validate — Making Evidence Verifiable
RealPeep
Zero-knowledge video provenance protocol. Validates that a video has not been tampered with between capture and viewing, using ZK proofs to verify the integrity chain without revealing the full video.
Sigstead
Cryptographic decision protocol. Validates that decisions were made by authorized parties at specific times, producing verifiable signatures and audit trails. Designed for governance and compliance contexts.
evidance
RAG (Retrieval-Augmented Generation) pipeline combined with cryptographic verification. Validates that LLM-generated responses are grounded in verified source documents. Retrieval + attestation.
📊 Surface — Making Evidence Actionable
graykm
Scientific writing and knowledge management. Transforms raw research data and analysis into well-structured scientific papers, technical reports, and documentation. Bridges the gap between computation and communication.
supps-lmn
Medical documentation pipeline. Surfaces structured patient records, supplement data, and lab result narratives from disparate sources. Built for accuracy and regulatory compliance in healthcare contexts.
Blog & Publications
Technical writing on evidence pipelines, ZK proofs, computer vision, and systems architecture. Making complex ideas accessible through clear, code-backed explanations.
Automation Tools
A suite of automation tools that surface results from complex workflows — CI/CD pipelines, data processing chains, deployment orchestrators, and monitoring dashboards. Turn processes into products.
Many of these projects are open source. Check out my GitHub for code, documentation, and ongoing work.
View on GitHub