Lead Data Scientist

About Consilio

Role overview

Responsibilities

• Own research workstreams from problem framing and scoping through experimentation, prototyping, and handoff to engineering and product.
• Set direction for implementing, fine-tuning, and deploying local LLMs for efficient eDiscovery analysis of legal documents such as emails and PDFs.
• Define legal-task evaluation methods and benchmarks and ensure results are reliable for decisions.
• Mentor data scientists through experimental design reviews, code review, and pairing; raise research rigor across the team.
• Evaluate emerging models, tools, and techniques and make build, buy, or adoption recommendations.
• Build rapid prototypes and judge whether they merit production development or should be set aside.
• Apply statistical analysis and traditional NLP alongside modern methods to enterprise and legal documents.
• Communicate findings, limitations, and trade-offs to nontechnical leaders and influence the product roadmap.
• Explore multimodal data and Knowledge Graphs to enrich solutions.

Qualifications

• 5–8 years of applied data science or machine learning experience, including modern NLP depth.
• Bachelor's degree in Computer Science, Data Science, or a related quantitative discipline; a master's degree is preferred.
• Strong background with LLMs, statistical analysis, traditional NLP, and enterprise documents, particularly legal documents.
• Hands-on experience with locally hosted and open-weight LLMs.
• Relevant tools include vLLM; Kubernetes and containerization; PyTorch; Hugging Face Transformers; PEFT/LoRA; LangGraph or comparable frameworks; embedding models and vector stores; spaCy; scikit-learn; DVC or similar; Docker; and Git.
• Research evidence of production impact, publication, patents, or other external recognition.
• Demonstrated experience mentoring or coaching technical practitioners.
• Multimodal data and Knowledge Graph experience is a nice to have.

Work mode

• Hybrid work mode requires three days per week in the office.