Building software that acts.

I work on systems that trade, hedge, route, reconcile, and recover without waiting for a human.

Currently building autonomous trading and risk infrastructure. Previously worked across AI agents, market infrastructure, and zero-knowledge systems.

I also research the mechanisms underneath markets: liquidation, solvency, funding, verification, and what changes when software becomes the market participant.

what i build

markets

trading, hedging, execution, reconciliation, and risk infrastructure

agents

software that can reason about state, make decisions, and take actions

developer infrastructure

tools that make complex codebases easier for people and coding agents to understand

how i build

  • + build for failure before optimizing the happy path
  • + make autonomous systems observable, constrained, and recoverable
  • + work close to the boundary between software decisions and real financial consequences
  • + use research when engineering intuition stops being enough
  • + prefer problems where the answer isn't obvious yet

tools

  • rust · go · python · typescript
  • fix · grpc · postgres · redis · kafka
  • aws · kubernetes · terraform
  • agents · llms · zero-knowledge

experience

  • ostium· senior backend engineer
    jan 2026 – present
  • subconscious ai· ai engineer
    aug 2025 – jan 2026
  • hustle fund· angel squad member
    oct 2024 – present
  • starkware· exploration software engineer
    sep 2024 – apr 2025
  • sizzle· advisor
    apr 2024 – aug 2025
  • nethermind· researcher
    may 2024 – aug 2024
  • hifi bridge· software engineer
    sep 2023 – feb 2024
  • block convey· technical lead
    jun 2023 – aug 2023
  • great lakes institute of management· visiting faculty - blockchain
    mar 2022 – jul 2022
  • opx ai· blockchain lead
    jun 2021 – jul 2022
  • vyom labs· data scientist
    jan 2020 – mar 2021

featured projects

repokg

2026

Generates an AI-ready knowledge graph of a codebase: module inventory, import graph, branch/PR state, contributor history, and CI/Docker/Helm/Make surface, rendered to KNOWLEDGE_GRAPH.md and machine-readable JSON.

developer tools, code intelligence, ai agents

repocanon

2026

Analyzes a repository once and generates repo-specific context for Codex, Claude Code, Copilot, and Cursor from a normalized deterministic project model.

developer tools, ai coding, static analysis

astra

2025

An experiment in what software looks like when agents stop answering questions and start coordinating actions.

agents, orchestration, runtime

yield pool watcher

2024

An autonomous agent watching markets across 16 networks and deciding where attention should go.

agent, defi, multi-chain

earlier projects

instantresearch

2023

LLMs as active research systems before that became an obvious interface.

Winner — Cohere Track, HACKGPT @ Cornell Tech

llm, research

ic3

2023

Proving you qualify to vote in a DAO without revealing who you are or what you voted for. Credentials attested from the open web, assembled into an identity.

Runner-up — IC3 Blockchain Camp @ Cornell Tech

deco, attested credentials, private voting

open source

flashbots/contender

#337enabled all benchmark scenarios to run via ci, improving test coverage and deployment confidence
#363added an admin reclaim-eth cli subcommand to improve ux for managing testnet eth
#324introduced logic to skip contract redeployment if bytecode already exists on-chain
#249implemented checks to validate on-chain code before deployment

PeerDB-io/peerdb

#3468extended support for replicating time64 data types in clickhouse, improving compatibility across data connectors

daydreamsai/daydreams

#244added support and usage examples for paradex and avnu integrations in the agentic trading framework
#12built yield pool watcher – a multi-chain agent for defi, tracking apy/tvl across 16 networks

elsewhere

I occasionally invest in and advise technical founders.

Mostly interested in people working on things that sound slightly unreasonable before they work.