The relevant training is inference under uncertainty — recovering a weak signal from a noisy channel, and knowing when you have not. Whether it was learned on markets is beside the point, and often a disadvantage.
Research direction · Proprietary data
Built the private-markets record the desk runs on: federal filings assembled and reconciled at scale — Form D, Form ADV, 13F and CMS ownership — covering more than half a million private issuers and the principals named on them. Operates the affiliated origination desk that keeps it current, which is why the private-markets data here is proprietary rather than licensed from a vendor every competitor also buys. Sets research priority, and decides what a finished answer has to clear before it leaves the desk.
Quantum information · Error correction
Quantum information researcher at Harvard. Published in Physical Review X on magic-state generation with finite block-length quantum LDPC codes, with M. D. Lukin; in PRX Quantum on non-Clifford gates via non-Abelian topological order; and in Physical Review B on multipartite entanglement in Heisenberg antiferromagnets. Prior work at Q-CTRL on machine-learned qubit layout selection and deterministic error mitigation. The transferable discipline is inference under noise: recovering a weak signal from a channel that is mostly error, and quantifying precisely how much confidence it supports.
Model architecture · Applied AI at scale
Co-founder and Head of AI at WOMBO, where he built and ran model architecture against consumer-scale inference load. Founder of OMEGA Labs. Read mathematical physics at Waterloo. Selected for AI Grant Batch 1 alongside Perplexity, Cursor and Replicate.
Distributed systems · Inference infrastructure
Built and scaled the infrastructure behind more than a billion generations at WOMBO — multi-threaded inference, backend throughput, distributed systems under consumer load. Earlier work on autonomous drone computer vision with Northrop Grumman. Owns the ingestion and inference layer the research runs on.
Process qualification · ML for design
Selected to MIT's Multiscale Materials Design programme, applying machine learning to design optimisation. Applications Engineering Manager at AON3D. Led development of flight-qualified parts for the International Space Station with NASA and the Canadian Space Agency. Materials engineering with aerospace at McGill; taught deep learning there.
Client coordination
Runs engagement operations: onboarding, delivery cadence and the administration around research handover. First point of contact for anything relating to a live engagement.
We hire for the habit of being rigorously wrong, then correcting.