VP, AI Model Research & Development

PowerToFly; Zug; Hybrid; CHF 180.000 - 240.000; Vollzeit

Benefits dieser Stelle

Hybrid Work Model

Flex My Way

Grow My Way

Mental Health Days

Tuition Reimbursement

Volunteer days

Zusammenfassung

Thomson Reuters Labs in Zug, with a hybrid setup including London, leads the model development and deployment for professional workflows, training and evaluating LLMs, and guides the roadmap and release decisions for real users.

The role requires leading a team of scientists and engineers, partnering across product, infrastructure, and editorial/SME teams, and aligning priorities with what customers need across CoCounsel, Westlaw and Practical Law.

Qualifikationen

• Program ownership: You have led a production LLM program end to end (data strategy through training, evaluation, and the release decision) for models that shipped to real users, and personally made or ratified the calls.
• Technical fluency: Enough command of post-training methods (SFT, preference optimization, RL), distributed training, and evaluation infrastructure to set direction, spot a training run going wrong, and be credible in the room with your own technical leads, without needing to run the experiments yourself.
• Track record: Shipped models, widely-used open-source contributions, or peer-reviewed publications at top research venues (e.g., NeurIPS, ICML, ACL, EMNLP) that speak for themselves.
• Leadership: Experience leading an organization of comparable size and shape (ideally with a manager/director layer beneath you), attracting and retaining senior ICs, and partnering across product, infrastructure, and domain experts.
• Delegation with accountability: Comfortable delegating hands-on post-training and data-strategy execution to your team, while retaining full accountability for the outcomes.

Aufgaben

• Own the model-development roadmap end to end (data strategy, training, evaluation, and the release decision) for models that ship to real users across CoCounsel, Westlaw, and Practical Law.
• Set the target for an online, agentic reinforcement-learning pipeline, training with subject-matter experts in the loop, and hold the work accountable to it.
• Ensure data strategy and evaluation approach are rigorous and tied to measurable outcomes.
• Enable new product capabilities to be built directly on Thomson LLMs, with demonstrated cost and/or quality advantages over third-party alternatives.
• Build and retain a senior bench of scientists and engineers, including hiring the next wave of senior IC and leadership talent.
• Partner across product, infrastructure, and editorial/SME teams to align model priorities with what customers actually need.
• Make the model-quality, safety, and cost tradeoffs the role is accountable for, and represent that judgment to executive stakeholders.

Kenntnisse

Program ownership

Technical fluency

Track record

Leadership

Delegation with accountability

Jobbeschreibung

Thomson Reuters Labs in Zug, with a hybrid setup including London, leads the model development and deployment for professional workflows, training and evaluating LLMs, and guides the roadmap and release decisions for real users.

The role requires leading a team of scientists and engineers, partnering across product, infrastructure, and editorial/SME teams, and aligning priorities with what customers need across CoCounsel, Westlaw and Practical Law.