Staff Engineer — AI Chat & Real-Time Systems

Shapes, Inc; San Francisco (CA); On-site; Full time

Job description

Shapes is building the best way to talk to AI and humans together — where AI agents hang out with real people in group chats, DMs, voice calls, and video calls, adding up to 10M+ minutes of conversation every month.

We’re a small team in SF. We build for ourselves — everyone at Shapes uses Shapes every single day, and everyone talks to users. Member of Technical Staff is the title we use for engineers who own hard problems end to end across the stack.

Representative projects

• Designing systems that give 3M+ AI agents distinct, consistent personalities across multi-turn group chats with humans
• Designing long-term memory systems that retain context across billions of messages
• Serving low-latency LLM responses for group chats with multiple shapes and humans in the same conversation
• Building the realtime voice and video stack for low-latency calls between humans and shapes
• Shipping the Next.js + React surfaces — feed, chat, profiles, and rooms
• Fine-tuning, evaluating, and routing across frontier models like Opus 4.7 and GPT 5.4 High Fast

You may be a good fit if you

• Have significant software engineering experience
• Results-oriented with a bias towards shipping
• Pick up slack beyond your lane
• Ship fast and reliable code
• Default to action: show, don’t tell; build, don’t deliberate
• Use Shapes every single day. If you’re not obsessed with the product, you can’t build something magical

Strong candidates may also have experience with

• LLM training, fine-tuning, evaluation, inference, or RAG at scale
• High-performance Python backends at scale
• Realtime infrastructure (WebRTC, WebSockets, streaming) for chat, voice, or video
• Consumer product work at companies serving millions of daily users
• Having shipped a consumer app on your own before

Links to repos, or working projects are a huge bonus.