Unsloth AI
Open Source Fine-tuning & Training of LLMs
Founding ML Engineer
About the role
We’re looking for someone who loves open-source and collaborating with the OSS community but also has deep understanding in how transformers and Unsloth works.
_MUST COMPLETE CHALLENGES BEFORE APPLYING: _https://x.com/danielhanchen/status/1891194528931209644
Must have:
- Deep knowledge and implementation experience with multi-GPU training techniques such as FSDP
- Important contributions to ML open-source projects (e.g., transformers, TRL, kernels, Unsloth) with reviewable code.
- Heavy experience working with ML frameworks like Unsloth, Transformers, TRL, and PyTorch.
- Familiarity with inference engines such as vLLM
- Familiarity with low-level languages and kernels like Triton
- Understanding of cloud services such as GCP or AWS (bonus).
Can be initially remote, with the ability to work in-person in San Francisco later.
About Unsloth AI
At Unsloth, we’re making custom AI as accessible as possible while reaching the widest possible audience. We're known for fixing critical bugs in LLMs like Gemma, Llama, and Phi-4, as well training bugs like Gradient Accumulation. If you want to work at a startup at the edge of innovation, you’re in the right place.
Based in San Francisco, Unsloth AI (YC S24) is backed by top-tier VCs and angels like Logan Kilpatrick (Lead Product at Google AI), Cliff Obrecht (Canva co-founder) and Jon Oringer (Shutterstock founder).
With 29K+ GitHub stars, 8M+ monthly downloads on Hugging Face, and adoption by Canva, LinkedIn, NASA, NVIDIA, and more, we’re shaping the future of AI. Join us and help build AI the way it was meant to be.
