Machine Learning Researcher

About the Company

A venture-backed AI company building 3D computer vision models that turn live video into spatial data. Early deployments are in live events and entertainment, where the output powers analytics and immersive media, with longer-term applications in robotics and embodied AI.

The team combines researchers and engineers from established tech companies and research labs, and the company has the data, compute, and funding to train its own foundation models.

The Role

This is a senior research leadership seat for someone who still writes code. You will partner with the CTO to decide what the company builds in 3D and 4D reconstruction, run the experiments yourself, and lead the researchers and infrastructure engineers who take that work into production. It suits someone who has already shipped reconstruction or neural rendering models into a live product and wants to set the research direction and standards for a growing team.

What You Will Do

Own the research roadmap for real-time, feed-forward 3D and 4D reconstruction, from early exploration to deployed models

Design model architectures, training approaches, evaluation protocols, and dataset strategy, and run the experiments hands-on

Lead and hire researchers and infra engineers, giving clear technical direction while leaving room for exploratory work

Review new academic and industry work and decide which ideas deserve investment

Publish papers, file patents, and give technical talks where it makes sense for the company

What You Bring

MSc or PhD in Computer Science, Computer Vision, Robotics, ML, or a related field, plus a history of leading researchers and engineers

A proven record taking 3D reconstruction, neural rendering, or spatial AI work from prototype into a production product

Deep knowledge of multi-view geometry, structure-from-motion, NeRFs, 3D/4D Gaussian Splatting, and feed-forward reconstruction, backed by papers at venues like CVPR, ICCV, ECCV, NeurIPS, or 3DV, or by patents or open-source work

Hands-on experience training models with PyTorch DDP or FSDP across multi-node GPU clusters in cloud or HPC environments, and managing multi-view video, LiDAR, and calibration data pipelines

Legal right to work in Canada and willingness to work onsite (sponsorship not available). Robotics, world models, diffusion, CUDA, or NVIDIA simulation tooling experience is a plus

Why This Role

You get large-scale 3D and 4D data, a pipeline to capture more from live deployments, and the compute to train foundation models from scratch. The research function is still being built, so you will set the technical bar and working style the team grows around, with direct ownership of high-impact projects and an early-team equity stake.