I've spent about ten years inside the CRM, billing, lead-to-cash, and data systems those teams depend on. I like working with the operators who run the process, putting agents on top of the tools they already use, and handing back a workflow they can audit and control.
AI teams often hire engineers who have never owned a HubSpot pipeline or a billing reconciliation. MarTech and RevOps teams often hire generalists who have never shipped an evaluated agent system.
I’ve spent about ten years on both sides. Lately most of my work has lived on the AI and internal-tools side.
The systems below come from that overlap: applied AI where it helps, operational workflows where it has to hold up, and the data structures and guardrails that keep automation usable once it leaves the whiteboard.
Nothing publishes until it clears the gate.
An automated post-production pipeline with AI-assisted gating before anything publishes.
One connector interface, every system in sync.
A multi-connector ETL platform syncing QuickBooks, Copper, Basecamp, and PandaDoc to PostgreSQL and Airtable behind a shared connector interface.
Every record accounted for on the way into one identity layer.
A PostgreSQL-staged pipeline that ingests Airtable records and resolves them into a deduplicated master-contact identity layer with reconciliation reporting.
Intermedia usage, billed straight into ConnectWise.
Full-stack billing automation syncing Intermedia usage data into ConnectWise agreements.
Read the case study
CSV in, enriched contacts out, at a fraction of the cost.
A contact-enrichment pipeline: a Brave + Gemini CLI and an OpenAI + Firecrawl web UI, with per-contact cost tracking.
Read the case study
Where scattered business data becomes client intelligence.
A Postgres-backed client-intelligence system that syncs Airtable records, normalizes identity fields, and surfaces a unified client view with source lineage.
Read the case study
Keeps your AI coding sessions alive and restarts the ones that hang.
A terminal session supervisor for AI coding assistants.
Read the case study
Run your dev ops from your phone.
A macOS daemon and CLI for mobile-first development ops.
Read the case studyYou can open all six right now. They run on real server-side backends, use synthetic sample data, carry honest labels, and do not need a login.
A four-method connector contract that cut a new integration from about 2,000 lines to about 50 at the entry point, plus the restraint it took to keep the interface small.
Meter · Billing automationHow preview-before-sync, ghost-addition detection, and DB-level locking help keep MSP billing reconciliation from duplicating or dropping charges.
I focus on retrieval accuracy, the guardrails that keep failure rates low, the observability that makes agent behavior diagnosable, and the handoff between agents and the systems they touch: billing platforms, CRMs, and internal tools.
I optimize for small surfaces, strong contracts, eval gates, and handoffs operators can live with day to day.
When a process is manual, fragile, or invisible, when data spans systems that do not agree, or when an AI workflow needs a quality bar it cannot fake.
Pure ML research, ground-up infrastructure or Kubernetes platform work, or anything that needs me to own visual design too. I'd rather flag a stretch up front than oversell the fit.
If you're building agent-assisted systems, internal tools, or the RevOps and MarTech backbone under them, feel free to reach out. I'm always up for comparing notes.