Doctolib Launches Clinical AI Lab to Build Patient World Models
balazskegl · x · 2026-08-05
European healthcare tech company Doctolib is hiring research scientists for its Clinical AI Research Lab in Paris. Directed by Balazs Kegl, the lab will begin reviewing applications in early September.
Core Research Focus
The lab aims to build a 'world model of a patient.' This model learns a representation of a person's health from their medical history to predict health evolution and response to care. It requires tackling cutting-edge problems like self-supervised learning over long, sparse, multimodal sequences, causal inference from observational data, certifiable calibration, and out-of-distribution evaluation.
Data & Deployment Advantages
- Massive Data: Doctolib is used by 570,000 health professionals and 90 million people across Europe, providing massive longitudinal ambulatory records to train personalized models.
- Closed-loop Feedback: Embedded within the product company, the lab can deploy models directly into daily software used by practitioners and patients, learning from real-world feedback to continuously improve care.
The lab is looking for early-career and experienced researchers with depth in areas like representation learning, LLMs, causal inference, world models, agentic systems, or privacy auditing. Experience with medical data is a plus but not required.
More from AGI Musings
- OpenAI's Safety Framework Under Fire: Gov Review 'Too Late' to Prevent Internal Leaks — ShakeelHashim · 2026-08-05
- AI4 2026 Kicks Off: Hinton and Ng Share Stage to Debate AI Governance — eyishazyer · 2026-08-05
- AI alignment debate: Is Claude pretending? RatOrthodox and So8res clash — sjgadler · 2026-08-05
- AI philosophy debate: Eliminativism wrong, natural science right? — AaronBergman18 · 2026-08-05
- New AI Career: 'Vibe Deployers' Earn Six Figures in 30 Days, Charging $3K-$10K per Setup — eyishazyer · 2026-08-05
- Karpathy Declares Gradient Descent Can Write Code Better Than You — RichmanRonald · 2026-08-05