Research Scientist, Interactive AI (Vision/Speech/Language)

Have you published, shipped, or both? Do you like designing interactive AI systems? Do you often build new architectures or data pipelines because existing models aren't good enough for the job?Then get in touch. Our team is in central London and partly in Amsterdam, so it's ideal if you're local or open to relocating.Though we are still in stealth mode for a few more months, we are actively hiring for multiple roles. We build human-centric AI for the next generation of policing. We build tools that help officers serve the public in a transparent, consensus-based way. To be clear, this is not mass-surveillance. Our mission is supporting the police on crime-investigation and helping victims, so the values of the people we hire matter to us.We are growing the applied-research side of our team because a) some breakthroughs are essential for the technology we're building and selling, and b) we believe in contributing to scientific progress through patents and (when appropriate) publications.We are working on a spectrum of fascinating applied research problems! Some are cloud- while others are edge-based. They challenge current AI capabilities across multiple domains:Speech: recognition and dialog systemsComputer vision: 3D reconstruction, recognition, grounding, VQALanguage: knowledge-grounded reasoning (RAG, neuro-symbolic)ResponsibilitiesScientific exploration of multimodal AI systemsEvaluation and benchmarking for interactive modelsBuilding data-collection processes and open-loop simulatorsDesigning and adapting efficient AI architecturesYour ability to absorb and apply new research can show up in a CV in many ways, e.g. publications, blogs, public code. Apply via LinkedIn, or here with a link to your homepage, Scholar, or GitHub: https://forms.gle/Q5399f6ZmxPLfUYf9Know someone you admire and trust? Please share this with them.