Researcher, Models (UK)

TL;DR

Researcher, Models (UK): Conduct groundbreaking research in neural network architecture design to advance the state-of-the-art in alternative architectures like state space models and efficient Transformers with an accent on improving model quality, inference efficiency, and adaptability. Focus on designing novel architectures for statefulness, long-range memory, and conditioning mechanisms while investigating trade-offs in scalability, robustness, latency, and energy efficiency.

Location: London, UK (Remote, in-person team based in London office)

Company

Pioneering model architectures for AI that learns from and interacts with the world like humans, founded by Stanford AI Lab PhDs who invented State Space Models.

What you will do

• Conduct groundbreaking research in neural network architecture design for state-of-the-art alternative architectures.
• Design novel architectures improving model quality, inference efficiency, and adaptability across deployment environments.
• Explore statefulness, long-range memory, and innovative conditioning mechanisms for model expressiveness.
• Investigate architectural impacts on scalability, robustness, latency, and energy efficiency trade-offs.
• Develop frameworks and tools to evaluate and benchmark architectural innovations.
• Collaborate with cross-functional teams to translate research into scalable real-world systems.

Requirements

• Based in London, UK with in-person office presence expected.
• Deep expertise in architecture design, including state space models, transformers, RNN/CNN variants.
• Strong understanding of architectures with system constraints like cloud or on-device deployment.
• Proficiency in balancing quality, efficiency, and adaptability across modalities (vision, audio, text).
• Familiarity with generative modeling (autoregressive, diffusion) and statefulness/conditioning.
• Proven research track record in top ML/AI venues (NeurIPS, ICML, ICLR, CVPR).
• Strong programming in PyTorch/TensorFlow and profiling tools.

Nice to have

• Prior research/publications in state space models or efficient Transformers.
• Experience in multi-modal systems architectures.
• Early-stage startup or rapid R&D innovation experience.

Culture & Benefits

• In-person team culture in San Francisco, London, and Bangalore offices; love collaborating daily.
• Competitive base salary with attractive equity.
• Commuter allowance, flexible PTO, daily meals and snacks.
• Visa sponsorship on case-by-case basis, dependent on role and location.
• Open, inclusive culture focused on support, resources, and fast execution without sacrificing quality.