Sorbet AI

AI, ML and computer vision consulting

Sorbet AI helps teams take models from research to production, from cloud training runs down to low-latency inference on edge devices.

Email us
Sorbet AI logo: a sundae glass with three scoops

Expertise

  • Computer vision

    Detection, segmentation, tracking, pose estimation and video understanding, trained on your data and measured on your cases.

  • Model optimization and edge deployment

    Quantization, pruning, distillation and compiled runtimes that make models fast and small enough for real-time use on embedded and mobile hardware.

  • Robotics

    Perception, state estimation and learned control for robots that have to work outside the lab.

  • 3D reconstruction

    Multi-view geometry, depth estimation, SLAM, NeRFs and Gaussian splatting.

  • Machine learning systems

    Custom models with the training, evaluation and deployment pipelines your own team can run and extend.

  • LLMs and agents

    Coding and research agents, retrieval, and the evaluations that show whether they actually work.

Built in the open

litledger

litledger is our self-hosted literature ledger for researchers and their coding agents: one shared library of papers, notes and evidence that Claude Code and Codex can read and write.

litledger on GitHub

Let's talk

Tell us what you're building and where it's stuck.

aidin@sorbet-ai.com

Vancouver, BC, Canada