AI Researcher

Kaedim; Greater London; On-site; GBP 70,000 - 120,000; Full time

Benefits offered by this job

Office in London

Fast-moving, impactful work

Job summary

Kaedim is seeking an AI Researcher to tackle open-ended problems at the frontier of what's possible. You'll translate breakthroughs into scalable approaches for the engineering team, shaping our AI pipeline and data infrastructure for next-gen robot training.

Requirements include an MS/PhD or equivalent, strong CV/3D AI expertise, a track record of publications, and fluency with Python and PyTorch. Offices are in multiple cities, with a collaborative, fast-moving environment.

Qualifications

• Advanced degree (MS/PhD) in Computer Science, ML or related field, or equivalent research experience.
• Track record of research contributions with publications at major venues.
• Deep expertise in CV, 3D generative models, 3D deep learning or robotics/embodied AI.
• Experience designing benchmarks, evaluation protocols and reproducible experiments.
• Proficiency in Python and PyTorch; ability to communicate findings clearly.

Responsibilities

• Lead original research moving pipeline forward in physics property prediction, 3D decomposition, scene reconstruction, and sim-to-real transfer.
• Design and run rigorous experiments with clear benchmarks and evaluation protocols.
• Author papers and technical reports communicating novel findings.
• Investigate hard, open problems across CV, generative 3D, and embodied AI and translate into roadmap.
• Collaborate with simulation and ML engineers to hand off validated approaches.
• Build and maintain evaluation benchmarks measuring pipeline quality against ground truth.

Skills

First principles research

Benchmark design

Strong writing & presenting

Python programming

Education

MS/PhD in CS/ML

Equivalent research experience

Tools

PyTorch

Python

Job description

As our AI Researcher, you'll work on the open-ended problems that sit at the frontier of what's currently possible and translate what works into approaches the engineering team can build on. The research agenda maps directly to the pipeline: every breakthrough here shows up in what we ship. This is a role for someone who can work from first principles on unsolved problems and move fluidly between scientific rigour and practical impact.

About Kaedim

Kaedim is an a16z-backed startup that raised $15M in our Series A to build the data infrastructure powering the next generation of robot training. Our AI pipeline turns 3D assets and real-world environments into validated, sim-ready datasets at the quality and scale robotics companies need to close the sim-to-real gap. We sit at the intersection of generative 3D, physics simulation, and physical AI, and we're just getting started. Explore what we're building at projectsim.ai

What you'll do

• Lead original research in the areas that directly move our pipeline forward — physics property prediction, 3D part decomposition and articulation, scene reconstruction, and sim-to-real transfer
• Design and run experiments rigorously, with clear benchmarks and evaluation protocols that let us measure progress honestly over time
• Author papers and technical reports that communicate novel findings to the broader research and industry community
• Investigate hard, open-ended problems across computer vision, generative 3D, and embodied AI, and bring relevant advances into our research agenda
• Work closely with simulation engineers and ML engineers to hand off validated approaches and help shape what gets built
• Build and maintain evaluation benchmarks that measure pipeline quality against ground truth

What you'll bring

• Advanced degree (MS/PhD) in Computer Science, Machine Learning, or a related field, or equivalent research experience
• Track record of research contributions — published papers at venues like CVPR, ICCV, NeurIPS, ICLR, SIGGRAPH, or CoRL are a meaningful signal
• Deep expertise in at least one of: computer vision, generative 3D models, 3D deep learning, or robotics/embodied AI
• Strong experience designing benchmarks, evaluation protocols, and reproducible experiments
• Proficiency in Python and ML frameworks such as PyTorch
• The ability to work from first principles on ambiguous, hard problems and communicate findings clearly to both technical and non-technical teammates
• Strong scientific writing and presentation skills

Why Kaedim?

• Work on a genuinely hard and important problem — the data bottleneck that's limiting physical AI
• a16z-backed Series A company with offices in London, New York, San Francisco, and Singapore
• Build the research foundation that the next generation of robotics companies will train on
• Small, technical team where your work ships fast and matters

How we work

We're a small, ambitious team that moves fast and takes real ownership of our work. Everyone is in the office five days a week, typically 9 to 9, because we're genuinely excited about what we're building and want to see it ship. If you want to work alongside people who care deeply, move quickly, and feel the direct impact of their work every day, we'd love to hear from you. You can read more about what it's like to be part of the team at github.com/Kaedim/working-at-kaedim.