Systemsthatship.
Machine learning engineer. I build production ML and agentic AI systems — models, the infrastructure they run on, and the agents that keep them honest.
Seattle, WA · Building in private
2×.
Offer acceptance · 11.5% → 26%
Event-driven.
A flexible backbone for AI systems
85%.
Of failures, fixed without a person
–40%.
Pipeline runtime · $95K/yr saved
+20%.
Auto-priced rate · patent pending
Δ Γ Θ ν.
Options, priced against your view
3 moments.
Found in the video, not designed
selected work
6 in detail- 2026agentic selling
A user-facing agent that writes the case for every quote — comparable research, LLM-judge selection, and a self-correcting generator producing a per-quote sell pitch. First phase more than doubled offer acceptance.
Built at Alt · LLM agents, Python, AWS
- 2026event-driven agent backbone
The SQS + Lambda event backbone that triggers the agentic systems — domain events in, agent workflows out, with retries and dead-letter queues instead of crons and crossed fingers.
Built at Alt · AWS SQS, Lambda, Terraform
- 2026pricing on-call agent
An on-call agent for pricing alerts. On a schedule the on-call sets, it sweeps the alerts that need work, investigates each across Airflow, Datadog, CloudWatch, SQS and GitHub, then restarts, reruns, opens a draft PR or escalates — and reports in the alert's own Slack thread. The guardrails live in code, not the prompt.
Built at Alt · Claude Code + MCP, Python, AWS
- 2025auto-pricing triage
Decides, item by item, whether a pricing model's price can be trusted: deterministic rules first, then evidence, then one route for each item: accept the price, send it to an agent pricer, or send it to an expert. Raised the auto-priced rate by more than 20%. Patent pending, named inventor.
Built at Alt · Patent pending · Python
- 2025ClickMoment
A tool for video creators that finds the frames already worth a thumbnail. It listens, watches, fuses the signals into scored segments, samples densely where they agree, and has Gemini name three moments. Shipped end to end on Cloud Run; now inactive.
Personal build · FastAPI, Gemini, Cloud Run, Supabase
- 2022ML platform optimization
The infrastructure every pricing model runs on — re-architected around Snowflake and Airflow for a 40% faster production pipeline, $95K/year lower compute spend, and deploys gated by the team's first pre-production testing framework.
Built at Alt · Snowflake, Airflow, AWS