Tractable
ML Intern · London, UK · Series E, $1B valuation
- • MLOps and research for automotive insurance at one of UK's top computer vision unicorns.
Work
Education
Yale University
B.S. Computer Science & Mathematics
Eton College
King's Scholar · Sixth Form Select · A-Levels (Math, Further Math, Physics, CS)
Experience
May 2026 – Present
Yale Graph & Geometric Learning Lab
Undergraduate Researcher · New Haven, CT
- • LLM architectures and post-training/RL for multimodal time-series reasoning.
- • Building data pipelines for next-gen foundation models.
SPAR Research
AI Alignment Research Fellow · Berkeley, CA
- • Activation steering in chain-of-thought LLM obfuscation with MIT and ex-Meta researchers.
ReefSound
Co-Founder · London, UK
- • Built audio AI & underwater robotics for coral reef monitoring with 99.4% accuracy using sound recordings.
- • Scaled product to 7 countries and 3 continents; news feature by The Independent, NASA, National Geographic.
- • Winner of NASA Conrad Challenge (2000 teams, 50 countries); won ~$300k in scholarships.
- • Researched novel CNNs + gradient boosting models; trained AI on 130+ hours of coral reef recordings.
Greyparrot
ML Engineer Intern · London, UK · Series B, $30M raised
- • Engineered PyTorch inference pipeline with test-time augmentation deployed across GPU clusters, improving F1-score while optimizing real-time processing speeds for complex computer vision tasks.
- • Built YOLOv8 object detection models to accurately classify 80+ distinct classes for industrial recycling facilities.
- • Worked directly with Head of Deep Learning and agile engineering teams to rapidly iterate and ship ML solutions.
CVSSP Lab, University of Surrey
ML Research Intern · Guildford, UK · #1 ranked computer vision lab in UK
- • Researched GAN model (CycleGAN + CLIP + InceptionV3) in PyTorch to perform complex image-to-image translations from full-color to flat-color; built computational photography pipeline for unpaired datasets.
- • Built novel sequence-to-sequence architectures (BART transformer + graph convolutional networks) for automated audio captioning; benchmarked on the Clotho dataset.
Technical
LanguagesPython · TypeScript · JavaScript · C++ · Swift · SQL · Bash
ML / AIPyTorch · TensorFlow · Scikit-Learn · HuggingFace · Pandas · NumPy · Stim
WebNext.js · React · Tailwind · Supabase · PostgreSQL · MySQL · Prisma