Jochem GietemaBuilding something new: speedrunning my way to solving big problems with robots and computer vision. Living in London, from the Netherlands.
Previously: led a team of ML Scientists at Onfido.
Research areas: vision-language model training, efficient and on-device ML.
Reading list
Favourites
GitHub
Send me some feedback!
Blog
- October '26 - Software 2.0 writes Software 1.0
- September '26 - What to build when robotics enters the smooth exponential
- August '26 - The future is robotics (data)
- April '26 - Insights by browsing 28 VLM eval datasets
- Loupy.ai - explore VLM eval datasets in your browser
- September '25 - Research notes: Reframing computer vision as token prediction
- May '25 - Coding Agent Frameworks
- May '25 - RL for long code sequences
- May '25 - Research notes: Mixture-of-Experts
- March '25 - Research notes: dataset generation for code LLMs
- February '25 - Biases in Apple's Image Playground
- January '25 - Research notes: Task vectors for vision
- December '24 - Research notes: few-shot performance of vision-language models
- May '24 - Research notes: scaling laws, emergent behaviour and inverse scaling
Before
- Developed a data visualisation tool Clusterfun, which is used at multiple leading computer vision companies in Europe.
- Co-authored a book about Bayesian Deep Learning - Enhancing Deep Learning with Bayesian Inference (Packt, 2023).
- Worked on document fraud detection at scale at Onfido, deployed to help verify the identity of more than 500K users a day (2020-2022)
- Worked with Dutch municipalities to build an online module for citizens to safe energy, helping more than 10K households to reduce their energy bills and carbon footprint (2017-2019)
- Created an early robot lawyer (2016)
- Studied Law (LLB, LLM) and Philosophy (BA) in Amsterdam (2011-2016)