Projects
Scalable microservices video platform with secure ingestion, automated transcoding, and global CDN delivery. Built on AWS EKS with RabbitMQ messaging, achieving sub-50ms latencies at 10k req/s. Optimized FFmpeg transcoding reducing processing time by 40%, and deployed CloudFront CDN cutting startup latency by 70%.
High-performance C++ PCAP parser processing IEX market data at 1M+ msgs/s with sub-200µs latency. Built modular ingestion system with SQLite bookkeeper for 1B+ record queries. Automated pipeline orchestration with Apache Airflow achieving 99.9% uptime and 3× throughput improvement.
Built a comprehensive AI context/memory layer for LLM agents with modular OpenAPI tool schema and real-time execution streaming. Implemented advanced prompt engineering and multi-agent coordination systems. Designed scalable architecture handling 10K+ concurrent AI conversations with sub-100ms response times.
An AI-first email client with a local LLM for maximum security, data sovereignty, and lightning-fast operations.
Built a complete interpreter in Go for a custom programming language, including lexer, parser, and evaluator. Implemented features such as first-class functions, closures, and a REPL.
Developed a real-time collaborative text editor with Yjs CRDT for conflict-free synchronization, custom WebSocket infrastructure, and AWS serverless deployment. Features include live cursors, document versioning, and multi-user editing with operational transforms.
Developed a fully functional Unix shell in C with support for piping, redirection, background processes, and custom built-in commands.
Implemented a custom malloc/free implementation in C that achieves performance comparable to the official published version. Optimized for both speed and memory usage.
High-throughput binary decoder for Nasdaq ITCH 5.0 protocol (650K msgs/sec, 1.68µs latency), with zero-copy parsing, profiling, and CSV export.