SDE 2 — AI / Full-Stack (GenAI, Computer Vision & 3D)

Job Description

About the role

You'll own significant pieces of AI systems built at the intersection of generative AI, computer vision, and 3D/spatial computing — turning complex visual and geometric inputs into production-grade output, shipped as real products. This is a high-autonomy, full-stack role spanning applied ML through deployed product, with room to shape how features are designed and built.

What you'll do

• Design, build, and ship generative-AI and computer-vision features end-to-end — owning them from prototype to production quality and latency bars.
  
• Work with 3D and geometric data — processing, analysis, and automated visual/spatial output.
  
• Own the full stack around these systems — model services, backend, and modern web frontends — including reliability, CI/CD, and production support.
  
• Break down ambiguous, research-flavored problems into shippable systems, and make sound technical trade-offs independently.
  
• Raise the bar through code review and by mentoring earlier-career engineers.
  

Requirements

Must-have

• Solid CS fundamentals plus 2–5 years shipping and maintaining production software.
  
• Hands-on Generative AI / applied ML — diffusion/vision models and/or LLM-based systems (RAG, agents) in production, not just notebooks.
  
• Strong core ML fundamentals — training, evaluation, and deploying models against real quality and latency constraints.
  
• Full-stack proficiency — a modern frontend framework plus a backend stack, with a track record of end-to-end ownership.
  

Good-to-have

• 3D / computer graphics / computational geometry background — meshes, spatial data, geometry processing, or rendering.
  
• Experience with generative video or conditioned/controllable generation.
  
• A record of research output or patents.
  
• Experience mentoring engineers.
  

You'll thrive here if you

• Move fluidly between a research idea, a working prototype, and a production fix.
  
• Like owning hard, open-ended problems end-to-end.
  
• Want breadth across ML, systems, and product rather than a narrow lane.