Systems Engineer focused on AI infrastructure, blockchain protocols, and verifiable computation.

Core stack: Rust, Solidity, Python.

My open-source work spans AI infrastructure, vector databases, agent frameworks, blockchain clients, protocol tooling, and distributed systems.

• [Merged] #2331 fix(langgraph): handle null thread checkpoint in RemoteGraph.getState in langchain-ai/langgraphjs
• [Open] #21336 fix(elasticsearch): split sync and async store paths in run-llama/llama_index
• [Open] #3996 fix(provider): poll receipts while waiting for confirmations in alloy-rs/alloy
• [Open] #24337 fix(download): avoid checksum scan during resume startup in paradigmxyz/reth
• [Open] #8957 Fix 8935 match except dev in qdrant/qdrant
• [Open] #6535 fix(ethereum): handle trace_filter traces missing result.output via c… in graphprotocol/graph-node
• [Open] #5461 fix(converter): fall back on invalid JSON-like partial matches in crewAIInc/crewAI
• [Open] #21386 fix(azureaisearch): preserve falsy metadata values in index mapping in run-llama/llama_index
• [Merged] #39169 fix(gdn): Align prefill warmup with real prefill path in vllm-project/vllm
• [Open] #10 Fix #9: update guest code for current nssa_core program API in logos-co/logos-lez-rln

ProofBoard is a protocol correctness workspace for smart contracts.

Problem: many protocol failures come from mismatched assumptions, unstated invariants, and weak verification workflows rather than obvious syntax bugs. Traditional audits are necessary but point-in-time, and they do not always encode evolving protocol intent as executable checks.

ProofBoard turns protocol intent into testable artifacts:

• explicit protocol assumptions
• executable invariants
• verification workflows tied to those invariants
• evidence ledgers for verification runs
• audit-readiness packets for review and handoff

Why this matters in DeFi:

• protocol safety depends on behavior under adversarial conditions
• correctness requires continuous validation, not one-time review
• teams need traceable evidence connecting intent to test outcomes

Long-term vision: make verification-oriented protocol development the default workflow from design to deployment.

Deterministic CI failure investigation for GitHub Actions with evidence-first RCA artifacts, guarded fix generation, and reproducible validation loops.

• AI Coding Agents Need Evidence, Not Confidence
• CI Failures Are Not Text Problems. They Are Execution Problems.

Writing focus:

• evaluation-first AI engineering
• deterministic agent workflows
• CI reliability and evidence-backed RCA
• guardrails and validation in AI-assisted development

• LLM orchestration
• Agent systems
• RAG / retrieval workflows
• Evaluation pipelines
• AI reliability
• LangGraph
• LangChain
• vLLM

• Rust
• Solidity
• Move
• Stellar

• GitHub
• LinkedIn
• Medium