Data infrastructure and AI systems consultant

I'm a data systems engineer who builds the infrastructure other systems depend on, from storage engines to real-time streaming. Over the last few months I've released three open-source projects, and I'm about to release a distributed agent runtime built upon them.

I'm available for independent consulting and contract work, building custom AI systems and data infrastructure for companies without a large software engineering team. To discuss a project, email [email protected].

github linkedin resume email

What I build

AI that does the work

A custom AI office for your business

A team of AI agents that keep working in the background and take on whatever workflow you need handled. They split up the work, hand tasks to each other, and keep a complete record of what they did.

Customer email, handled

Reads every incoming message, sorts it by urgency, and drafts a reply for a person to approve.

Paperwork, processed

Reads invoices, contracts and forms, pulls out the key details, files them, and flags anything unusual.

Leads, followed up

Researches each new inquiry and drafts the first reply, so no lead waits overnight.

A morning briefing

A plain-English summary of what changed across your tools and what needs your attention today.

An assistant that knows your company

Answers staff questions from your own documents and shows where each answer came from.

Data infrastructure

All your data in one place

Pulls data out of your separate tools into one organized, searchable store, so reports and AI projects stop starting with a spreadsheet hunt.

Live data you can trust

Captures events as they happen (orders, sensors, usage) and stores them durably, so nothing is lost and everything can be replayed.

Fixing slow or fragile systems

Speeds up, simplifies and lowers the cost of data pipelines and databases that have become hard to run.

Reports and dashboards that stay correct

Built on the same data your systems use, so the numbers agree.

Custom projects, by request

If your problem isn't listed, tell me what's slowing your team down.

Open source

agent-officeopen-sourcing 2026.11

Python · Claude Agent SDK · TypeSafe SDK · LiteLLM

A runtime for agent organizations: long-running agents, the services around them, and one durable message log between them.

streamcast

Python · WebSocket · JSON Schema · msgspec

A JSON WebSocket pubsub framework for structured data.

litelink

Python · Iceberg · Parquet · SQLite · DuckDB

An embedded Iceberg storage engine for append-only data.

penca

Rust · Flight SQL · DataFusion · Lance · Parquet

Branchable, versioned OLTP and OLAP on one open columnar copy of your data. An experimental proof of concept.

Writing

2026-05-19

Designing Penca: the version-controlled lakebase

A technical brief on how Penca collapses the production and analytics split into one version-controlled lakebase, covering the architecture, branching, row-level history, and the open standards it speaks.