Top Tech Jobs & Startup Jobs in San Francisco Bay Area, CA

Lead multimodal research strategy for video and sensor understanding at petabyte scale. Select models, representations, training approaches, and evaluation methods; transition prototypes into production inference; define benchmarks; and optimize distillation, routing, cascades, quantization, GPU utilization, throughput, and cost. Translate customer research needs into technical programs while remaining hands-on with papers, models, experiments, and architecture. This is a senior individual contributor role with no direct reports.

Owns Eventual’s operating cadence, including planning, goals, metrics, board and investor materials, and company decisions. Leads customer and partner conversations, contracts, procurement, security reviews, go-to-market initiatives, pricing, hiring, and ambiguous cross-functional projects. This generalist role moves across operations, legal, sales, marketing, and strategy while building scalable systems and eventually developing functions into standalone teams.

Build core multimodal backend systems for robotics and Physical AI, including real-time WebRTC video pipelines, hardware-accelerated codecs, petabyte-scale ingestion and storage, LiDAR and telemetry processing, sensor synchronization, and high-performance React and TypeScript interfaces. Engineers may work across infrastructure, spatial data, video, frontend, and AI-native product workflows while independently solving complex technical challenges.

Build distributed data systems for petabytes of video, lidar, and telemetry data. Responsibilities include developing multimodal data lake storage, query engines with indexing and vector search, and high-throughput dataloading pipelines delivering data to GPUs. The role involves optimizing distributed storage and compute, query planning, streaming computation, zero-copy data flow, and real-time indexing for robotics and Physical AI applications.

As a Research Engineer, you'll enhance visual understanding capabilities by training and deploying models to make video datasets queryable, driving down costs, and collaborating on data taxonomy design.