Reka

Reka builds multimodal AI systems designed to perceive, reason and act across video, images, audio and text. The company's stated mission is twofold: to model the physical world through multimodal understanding, and to put useful multimodal AI into the hands of people. Its models and systems are built for use by enterprises and organisations, and the company reports that its work reaches such users worldwide.

The technical scope is broad. Reka works across multimodal understanding, video, image and audio comprehension, natural language processing, physical world modelling, GPU performance engineering and AI systems research. Roles reflect that range, spanning research and infrastructure positions such as AI systems researchers, GPU performance engineers and AI tutors.

The team is drawn from around the world and operates remote-first, with members spread across a distributed, globally recruited unit. Researchers and engineers on the team have contributed to breakthroughs at Google DeepMind, Facebook AI Research (FAIR) and other leading AI labs. Working practices emphasise end-to-end ownership of complex problems, collaboration and openness, and cross-disciplinary work in which people frequently cover more than one role.