ML training lead

need a senior ML lead who has done the real thing: • Fine-tuning LLMs (LoRA, QLoRA, PEFT, full fine-tuning) • Reads training curves, designs ablations, and can explain why a result holds up • NOT prompt engineering or agent-building on top of frontier models — that’s a valuable but different skill • Model evaluation, benchmarking, and ablation studies • Preference optimization (DPO/RLHF) • Synthetic data generation and data curation • Agentic AI workflows and reasoning models • Distributed training and GPU optimization • Working with models such as Llama, Qwen, DeepSeek, Mistral, Gemma, and similar open models • Building repeatable experimentation pipelines and evaluation loops Bonus experience includes: * Knowledge distillation * RAG architectures * Multimodal models * Quantization and model optimization * vLLM, TensorRT-LLM, Ray, DeepSpeed, FSDP * Kubernetes and large-scale inference * Education domain experience would be a big differentiator. Structure: fractional / advisory to start, with the option to grow into the role as the work scales.