trace
span
topaz labs(dallas, tx)2y 8msenior ml platform engineer1y 5mml platform engineer1y 3m
- place
- dallas, tx
- when
- jan 2024 – now
- kind
- work
- systems
- go · kubernetes · nats · tensorrt
I joined to build the company's first inference platform. It has since become the system that schedules, runs and bills for GPU work across several clouds, and most of what I write about comes from here.
- 14 aug 2026post-mortemThe night Spot capacity disappeared
- 12 jun 2026decisionTwo-phase commit for credits, not sagas
- 09 apr 2026noteWhat 100 TensorRT conversions taught me about ONNX
- 06 feb 2026decisionRouting GPU work by complexity instead of queue depth
- 04 dec 2025noteNotes from mentoring two engineers
- 02 nov 2025post-mortemA NATS consumer that acknowledged too early
- 18 aug 2025decisionWhy the platform runs on Karpenter
troutwood(pittsburgh, pa)3m
- place
- pittsburgh, pa
- when
- may 2023 – aug 2023
- kind
- work
- systems
- python · aws lambda · plaid
A summer of serverless backends on AWS Lambda, wired to Plaid so that financial models could update as accounts changed.
carnegie mellon(pittsburgh, pa)1y 3m
- place
- pittsburgh, pa
- when
- sep 2022 – dec 2023
- kind
- study
- systems
- scalable systems · formal methods
The Master of Software Engineering, in the scalable systems track. I also taught Product Management Essentials and Formal Methods.
building blocks(tilburg, nl)2ysoftware engineer1y 10mintern2m
- place
- tilburg, nl
- when
- jul 2020 – jul 2022
- kind
- work
- systems
- angular · spring boot · scala
My first engineering job, which started as an internship. I built micro-frontends, Spring Boot services and a system for moving large datasets to clients.
tu delft(delft, nl)2y 9m
- place
- delft, nl
- when
- sep 2019 – jun 2022
- kind
- study
- systems
- reinforcement learning · finance
Computer Science and Engineering with a minor in finance, finished cum laude. My thesis was a negotiating agent trained with reinforcement learning.