About

What I build

I design and ship deterministic options research infrastructure. My work turns vendor REST data into replayable backtests that investment teams can audit inside their own environment.

Mission

Make vendor REST options data backtestable, reproducible, and auditable without standing up a data lake.

Why I built Vol Street

Across client work I kept seeing the same pattern:

• Teams used Amberdata, IVolatility, or similar REST providers but did not keep a database.
• NNBO is one day per request, so naïve hydration means hundreds to thousands of calls per symbol.
• Feature-only notebooks could not choose strikes from the chain, manage rolls, or enforce capital rules.
• Replays drifted because corrections and late prints were not versioned.

I built the core engine to solve those problems and deployed it for multiple clients. That engine is now Vol Street: a compiled Python SDK (Cython) that runs in the client's VPC, respects vendor limits, merges corrections with diffs, selects strikes from the live chain per policy, and exports signed artifacts tied to a run hash and seed.

What I do today

• Provider-agnostic backtesting systems that hydrate daily REST windows safely and reproducibly.
• Client data integration with licensed vendors and custom datasets without redistribution.
• Audit-ready results for investment committees and diligence, including manifests, notebooks, and signed PDF reports.

I package this as an SDK that institutional teams can deploy quickly with integration support for their environment.

Earlier work that shaped Vol Street

• Built trading systems and desk tooling: position tracking, greeks engines, and the data plumbing that keeps them stable at scale.
• Contributed to derivatives research at Amberdata after its acquisition of Genesis Volatility.
• Longstanding focus on event-driven backtesting, walk-forward validation, and ML pipelines for options and volatility.

Principles I will not compromise

• Determinism over demo – every run emits a manifest with seed, run hash, image digest, timezone, and library versions.
• Chain state at decision time – strike selection comes from the tradable chain snapshot, not a single delta column.
• Lifecycle realism – rolls, early exits, assignment, calendars, and fees are encoded as policy.
• Execution discipline – NBBO, size checks, and slippage rules gate fills.
• Client custody – runs happen inside your environment, with node-locked licensing and no telemetry.

How I work with funds

• Pilot – fixed scope proof on your vendor key that ends with a signed report, notebook, parquet and csv, plus a manifest. Pilot fee credits to production.
• Production license – perpetual right to run the compiled SDK in your environment, with implementation and handoff.
• Support and updates – vendor schema compatibility, bug fixes, minor features, and replay drift assistance.

If you are evaluating options infrastructure

If you want deterministic backtests on your own vendor data without building a platform, Vol Street gives you the engine and the guardrails. Bring your license. Keep your data. Get results you can replay.