Manager, Data Science
Core
Lead the design and execution of LLM training and fine-tuning initiatives, overseeing dataset preparation and establishing evaluation frameworks for model quality.
Role type
Manager, Data Science (LLM Training & Fine-tuning)
Builds
Production-ready Large Language Models (LLMs) for legal, regulatory, and business information analytics.
Domain
Legal technology / Natural Language Processing (NLP)
Deliverable
production ML models
Required skills
Transformer architectures, attention mechanisms, tokenization, language modeling, supervised fine-tuning, parameter-efficient fine-tuning (LoRA/QLoRA), multi-GPU training (PyTorch FSDP/DeepSpeed), experiment tracking, dataset versioning, benchmark design, human evaluation, error diagnosis, GPU memory optimization, mentorship
Preferred skills
None stated
Technologies
PyTorch, Python, Hugging Face Transformers, Datasets, LoRA, QLoRA, PyTorch FSDP, DeepSpeed
Responsibilities
Lead design and execution of LLM training and fine-tuning initiatives; oversee preparation of high-quality training datasets (collection, cleansing, deduplication, annotation); develop and improve fine-tuning workflows and preference optimization methods; establish evaluation frameworks for factual accuracy and safety; diagnose training issues and enhance model quality/stability; manage and mentor data scientists; partner with product and engineering teams; manage project priorities, timelines, and compute resources.
Seniority
Manager, hands-on IC with team leadership
