I'm the founding engineer at Wholesum, building systems to help understand qualitative data at scale.
Outside of work I like building side projects and learning new things. I also spend far too much time playing guitar and losing at chess.
Writing
Work
Wholesum
Founding Engineer
Feb 2026 - Present
- Building systems to help understand qualitative data at scale
- Working across infrastructure, inference and product with a small team
- Learning a lot about LLMs, evaluation and production AI systems
Malted
Software Engineer
Jun 2025 - Feb 2026
- Led backend feature development for the core product
- Built a conditional inference engine and Python DSL for routing between models
- Worked across deployment, fine-tuning and on-prem inference infrastructure
HackerNoon
Software Engineer
Apr 2021 - Feb 2024
- Migrated the platform to SvelteKit and Postgres
- Built inference infrastructure on GCP and Kubernetes used by millions
- Added features across the editor, comments, notifications and Web3 platform
Projects
- Code
AutoTab
Guitar tablature transcriber
- Adapted Whisper to generate guitar tablature from recordings
- Scraped and trained on a dataset of 100k+ community guitar tabs
- Reached around 85% transcription accuracy
- Code
Blunder
Cheat detection system for online chess.
- Trained a custom ViViT model to detect selective cheating
- Improved on the previous state of the art by around 14%
- Later adopted by Chess.com to support fair play investigations
- Code
PintFind
A tool to find the best pubs in your area
- Built a local pub discovery app with reviews and recommendations
- Grew to 500+ active users and 300+ reviews
- Later acquired by a local competitor
Synapse
Educational autograd engine
- Built a hackable autograd engine in Python and Rust
- Implemented SGD, Adam and RMSProp from scratch
- Trained a CNN on MNIST to 99% accuracy