dvrp
San Francisco, CA; full-time; onsite/in-person
I’m the cofounder of KREA (https://krea.ai), a startup in San Francisco building a design tool to make AI fast, controllable, and intuitive. We’re a small team (~7) and we’re looking for talented folks (https://www.krea.ai/careers) who are interested in scuffling hard technical problems with other smart people while building a useful, beautiful, and performant product that millions of creatives globally will rely upon.
Check out our SIGGRAPH talk (https://youtu.be/Gm1B5DT8kE0?t=5824) or how our users use KREA (https://x.com/MartinNebelong/status/1724747362327781502) to see some features of the product.
We are well-funded, have millions in both users and revenue, and we keep growing. Our userbase ranges from the creative director of Coldplay to people doing 3D renders for architecture firms.
However, everything is on fire, and our small team can't build fast enough alone. We are looking for any smart person fascinated by our industry.
Example tasks:
Frontend: building out 3D environments for AI rendering, highly detailed optimization work on our real-time AI canvas, and tons of Svelte (compiler knowledge is a great bonus!).
Machine-learning: work on adapters from MMDit blocks to make models like Flux or SD3.5 more steerable; apply SAEs to cluster and interpret large image/video datasets; implement Diffusion-DPO; work on large-scale training and inference optimization for image/video diffusion models.
Backend: build, scale, and "reliabilize" an infrastructure to support a cluster of >1000 GPU-nodes performantly, simply, and beautifully. We expect excellent understanding of distributed systems and engineering trade-offs.
Our tech stack: PyTorch, CUDA, DeepSpeed, k8s, ONNX, TensorRT, Grafana, Svelte(Kit), PostgreSQL, Redis.
If you are interested in learning more about what we’re working on or just want to meet to talk about steerable AI for creatives, you can email me at d+hn@krea.ai.