DATA SCIENCE
Foundation Model Engineer
Go below the API. You will adapt open-weight foundation models for clients who need control — continued pretraining, fine-tuning, distillation and the serving stack that makes a private model economical.
- Dallas
- Data Science
- Full-time
What you’ll do
- Adapt and evaluate open-weight models for client domains, languages and latency envelopes
- Run training infrastructure with disciplined experiment tracking and reproducibility
- Distil frontier-model behaviour into smaller models where cost or privacy demands it
- Design serving architectures that hit client cost-per-token targets
- Advise clients on model strategy: open weights, frontier APIs or hybrid
What we’re looking for
- MS or PhD in Computer Science, AI, or equivalent experience
- Deep PyTorch experience with distributed training (FSDP, DeepSpeed) on multi-GPU clusters
- Hands-on fine-tuning of open-weight LLMs (Llama, Mistral, Qwen) with LoRA/PEFT, SFT and preference optimisation
- Experience with quantisation and efficient inference (vLLM, TensorRT-LLM, GGUF)
- Judgement on build-vs-API trade-offs and the honesty to recommend the API when it wins
What we offer
- Competitive compensation package
- Health, dental, and vision insurance
- Dedicated GPU cluster budget
- Flexible work arrangements
- Access to cutting-edge computing resources