Are you a researcher who wants to do serious machine learning in a domain that matters? We are seeking an exceptional Postdoctoral Research Associate to join MindCraft, an innovative programme at Imperial College London developing AI-guided smartphone-based mental health interventions for adolescents. You will work at the intersection of machine learning, computational psychiatry, digital phenotyping, and real-world intervention research — building mathematically grounded, personalised AI systems that can infer mental state from multimodal behavioural data and learn to deliver the right intervention at the right time.
You will play a central role in developing the project’s computational and machine learning core. Your work will include:
1. Developing multimodal models of adolescent mental state from longitudinal mobile and self-report data
2. Designing latent-state, state-space, probabilistic, or representation-learning approaches for modelling mental health trajectories
3. Building personalised digital twin models integrating behavioural, contextual, and questionnaire-derived information
4. Developing reinforcement learning, contextual bandit, or sequential decision-making models for adaptive intervention delivery
5. Tackling core challenges such as partial observability, uncertainty, missingness, delayed rewards, and non-stationarity
6. Contributing to the prospective deployment and evaluation of AI-driven interventions within a real-world school-based study
Depending on your strengths, the role may lean more heavily toward probabilistic and dynamical modelling, reinforcement learning, multimodal representation learning, or computational behavioural modelling.
We are especially interested in candidates who combine mathematical maturity, serious implementation ability, and scientific curiosity about mind and behaviour. You should hold, or be close to completing, a PhD in machine learning, computer science, computational neuroscience, computational psychiatry, applied mathematics, statistics, engineering, physics, or a related quantitative discipline.
You should have expertise in one or more of: probabilistic modelling, time-series modelling, latent-variable models, state-space models, reinforcement learning, sequential decision-making, representation learning, Bayesian methods, or computational models of behaviour. Strong Python and deep learning framework skills (PyTorch, JAX, or TensorFlow) are essential. Experience with healthcare, mental health, or mobile sensing data is highly desirable but not required.
7. The opportunity to work on hard machine learning problems with genuine scientific depth and real-world clinical consequence, not just pipeline engineering.
8. Access to rich longitudinal data from an ongoing adolescent mental health study involving 700+ participants, with a rare opportunity to move beyond offline modelling into prospective, real-world adaptive intervention.
9. A highly interdisciplinary research environment working closely with clinicians, psychiatrists, and behavioural scientists, with a clear route toward meaningful clinical impact for adolescent mental health.
10. Sector-leading salary and remuneration package (including 41 days off a year and generous pension schemes).
11. Be part of a diverse, inclusive and collaborative work culture with various and resources to support your personal and professional .