Posted 16 days ago
Description
Founding Machine Learning Research Scientist / Engineer - Optimisation & Bayesian Deep Learning
London - Hybrid (Flexible, primarily Remote)
I'm hiring for an AI start-up focused on continual learning. They're looking for a Research Scientist/ Engineer to work on the mathematical and algorithmic foundations of this programme, combining advanced optimisation, probabilistic modelling and large-scale deep-learning experimentation.
You will join a small, highly technical group and take ownership of research from theoretical development through to implementation and empirical validation.
What you’ll work on
- Develop new second-order and curvature-aware optimisation methods.
- Research scalable approximate Bayesian inference for neural networks.
- Work with Fisher, Hessian and related curvature approximations.
- Develop uncertainty-estimation and posterior-approximation techniques.
- Investigate optimiser design, training dynamics and model adaptation.
- Implement new learning algorithms in modern deep-learning frameworks.
- Evaluate new methods across increasingly large neural networks.
- Help shape the company’s wider research direction.
What we’re looking for
- Deep expertise in optimisation, Bayesian deep learning or a closely related area.
- Experience with natural gradients, KFAC, Shampoo, SOAP or similar techniques.
- Knowledge of approximate inference and uncertainty estimation.
- Strong mathematical and numerical understanding of deep learning.
- Excellent Python and modern ML framework experience.
- Ability to translate research ideas into reliable experimental code.
- A PhD or equivalent research experience in a highly technical discipline.
Useful additional experience
- Training or adapting large language models.
- Continual learning, active learning or reinforcement learning.
- Distributed multi-GPU or multi-node training.
- Optimising memory, communication or optimiser state.
- Scaling novel algorithms from small experiments to large models.
Depth in optimisation and probabilistic machine learning is more important than experience with any particular model family.
- Work on fundamental, unsolved problems in machine learning.
- Own research from mathematical development through to large-scale experimentation.
- Join a small, high-calibre team with significant technical autonomy.
- Help shape the research culture of an ambitious early-stage AI company.
- Receive a highly competitive salary and meaningful equity.
#J-18808-Ljbffr
