AI Model Fine-Tuning Engineer

Take open and frontier models past their defaults - fine-tuning, distillation and alignment for domain-specific intelligence. WHAT YOULL DO Fine-tune and adapt LLMs and speech/vision models (LoRA/QLoRA, SFT, DPO/RLHF) for client domains. Build data pipelines: curation, labelling workflows, synthetic data generation and quality gates. Distil large models into fast, deployable ones without losing the behaviour that matters. Own training infrastructure - GPUs, experiment tracking, reproducibility. Benchmark rigorously against real business metrics, not leaderboard scores. WHAT WE LOOK FOR Solid ML fundamentals plus PyTorch experience; you have actually trained models, not just called APIs. Comfort with distributed training and inference optimisation (quantisation, batching). Scientific honesty - you report what the eval says, not what the demo shows.