Location: San Francisco Bay Area (On-site)
Salary: $130K to $300K
Equity: 0.5% to 2% depending on seniority
Work Policy: On-site 5 days a week. No remote. This is a collaborative hands-on environment
Visa Sponsorship: Available
We’re not looking for academics. We’re looking for engineers who deploy. If you’re deep in the stack, building models, wrangling data, working with hardware, and making it all run in the field, you’ll fit right in
This is not a research role. It’s a production-focused position where your code hits real world constraints, edge compute, real time requirements, unreliable environments, and still needs to work
What You’ll Be Doing
• Building and shipping CV models for action recognition and real world video analysis
• Training and deploying large scale models using production ML workflows
• Working directly with physical environments including edge compute, robotics, manufacturing, and construction
• Owning model lifecycle from data and annotation workflows to real time inference at the edge
What You Bring
• 3 plus years of hands-on experience in CV, ML, or related fields
• Strong background in PyTorch, distributed training, and live model serving
• Deep expertise in modern CV architectures including VLMs, CNNs, LSTMs
• Comfortable with hardware and software integration and performance tuning at the edge
• Experience with large annotated video datasets and multimodal AI (vision plus language plus time)
• MS or PhD in CS, EE, or a closely related discipline
Strong Plus If You’ve Worked In
• Robotics, autonomous vehicles, edge AI, or construction and manufacturing tech
• Field-deployed CV systems in production settings, not just simulated environments
Not a Fit If…
• You only manage teams and no longer write code
• You’ve only worked in academic or lab settings
• You’re disconnected from the mission or impact of applied engineering
This is a rare role. Deep tech. Real world. High ownership. You’ll be joining a team that moves fast, builds clean, and ships code that survives in the wild
Apply now if you’re the type of engineer who cares more about deployment than demos