One engineer. Five domains. Every layer of the stack.
AI systems, embedded hardware, blockchain, distributed infrastructure, and real-time streaming — designed under real constraints, shipped, and documented where they break.
Constraints drive the design. Failures ship with the docs.
Every system here was built against a hard constraint — memory limits, latency budgets, consensus overhead. The architecture decisions are documented alongside the dead ends, not just the version that worked.
Research methodology, integration decisions, and failure analysis carry equal weight to the code. That is the differentiator — reproducible patterns across domains, not isolated demos.
Where the shipped work lives
Neural inference at the edge
Consensus under adversarial load
Data pipelines that hold under pressure
PyTorch, TensorFlow, LangChain, PydanticAI, LLM fine-tuning, RAG pipelines, FreeRTOS, ESP32, MQTT, and real-time sensor fusion.
Solidity, Rust smart contracts, Bitcoin protocol internals, distributed consensus, vector databases, and fault-tolerant system design.
Kafka, Docker, Kubernetes, Grafana, FastAPI, cloud infrastructure, DevOps automation, and real-time analytics at production scale.
If you've read the work, the next step is direct.
Open to collaboration on systems that cross domain boundaries — AI-embedded integration, distributed ledger infrastructure, real-time ML pipelines. No recruiters; no cold pitches.