We're building a self-service LoRA / full fine-tuning platform for LLMs, deployed on AKS. This role owns the training pipeline and data pipeline — turning raw datasets into validated training inputs and running LoRA/full fine-tuning jobs on GPU infrastructure.
What you'll do
- Build and improve LoRA and full fine-tuning workflows (PyTorch + PEFT + CUDA)
- Own dataset ingestion, transformation, and validation
- Work within existing config and logging contracts shared with other platform teams
What you need
- Strong Python and hands-on PyTorch experience
- Experience with LoRA or full fine-tuning of LLMs
- Comfort debugging CUDA/GPU training workloads
- Experience building data ingestion/validation pipelines
Nice to have
- vLLM or LLM inference-serving experience
- Azure (Blob Storage, Snowflake) experience
- Basic Kubernetes familiarity
From the posting on Jade Global's job site. Read the original before you apply.