I'm a Member of Technical Staff at World Labs, working on spatial AI. My team from Coplay was acqui-hired here in early 2026. World Labs is building world models and was founded by Fei-Fei Li.
I was Co-Founder and CEO at Coplay (acquired). We built an agentic AI coding assistant for Unity game developers -- think Cursor for game engines. We grew to 8K monthly active users and 8K GitHub stars.
The hardest problem was bridging the gap between "LLM can generate code" and "developer actually trusts the output." We designed a DSL called Scene Language that grounded LLM output in structured, verifiable actions, reducing token usage by 90% and making the AI's work legible. I led a team of 8 engineers shipping weekly releases.
I was Co-Founder and CEO at Remio. We raised $4.5M from Khosla Ventures and built a VR collaboration platform for enterprise, serving 30+ Fortune 1000 companies including Google, Netflix, and HubSpot.
We pivoted from enterprise to consumer and built Animal Blocks, a VR game similar to Minecraft, which grew to 150K monthly active users and profitability. I built and led a team of 17 across engineering, art, and product.
I was Co-Founder and CEO at Kristalic. We built AI-powered meeting summarization and action item extraction -- essentially a pre-cursor to Otter.ai. We designed an end-to-end ML pipeline with BERT models and SincNets, and built an RLHF-style feedback loop that reduced user downvotes by 83%.
My PhD at Cambridge University was focused on the Long-Short-Term-Memory (LSTM) neural network model in the context of biomedical time-series data (ECG, EEG, glucose, etc.). I created a simplified LSTM architecture (JANET) that achieved state-of-the-art performance with 50% fewer parameters, published at ICLR and NeurIPS workshops.
I was part of the Signal Processing and Information Engineering group at the Engineering Department. Here I was fortunate to overlap with Zoubin Ghahramani and Carl Rasmussen from the CBL group, and Alex Kendall who founded Wayve at the desk next to me.
I finished my PhD in 2.5 years (vs typical 4) while consulting for 3 ML startups on the side -- including BIOS and Cambridge Cancer Genomics (YC). That's where the startup bug bit me.
Started my MEng in Biomedical Engineering at Stellenbosch University. Worked on Self-Organizing Maps for EEG classification. I.e. machine learning applied to brain signals. This is where I first got into machine learning. I dropped out to start my PhD at Cambridge.
BEng in Mechatronic Engineering at Stellenbosch University. This is where I got into biomedical engineering and robotics -- the team I led won the national Pneudrive Challenge in 2014 with an autonomous pallet stacking robot. I graduated top of my class with a Cum Laude.