Vice President, Model Research & Development

Thomson Reuters Labs, the global R&D arm of Thomson Reuters, builds cutting‑edge AI solutions that help professionals make faster, better decisions. As Vice President of Model Research & Development you will steer the team that creates fiduciary‑grade large language models (LLMs) used in products such as CoCounsel, Westlaw and Practical Law. In this role you will own the full model‑development lifecycle: define data strategy, oversee training, run rigorous evaluation, and decide when a model is ready for release. You will set targets for an online, agentic reinforcement‑learning pipeline that incorporates subject‑matter‑expert feedback, and ensure that data and evaluation methods are tightly linked to measurable outcomes. Collaboration is key, you will partner with product, infrastructure and editorial experts to align model priorities with real customer needs, while making trade‑offs between quality, safety and cost and presenting those judgments to senior executives. Building and retaining a senior bench of scientists and engineers, including hiring the next wave of senior individual contributors and leaders, is also part of the daily agenda. The position requires a proven record of leading a production LLM program from data strategy through training, evaluation and release, with personal responsibility for the final decisions. You must possess deep technical fluency in post‑training techniques (such as supervised fine‑tuning, preference optimization and reinforcement learning), distributed training systems and evaluation infrastructure, enough to guide teams and spot issues without running every experiment yourself. A strong track record of shipped models, widely‑used open‑source contributions or peer‑reviewed publications at top venues (NeurIPS, ICML, ACL, EMNLP) is essential. Leadership experience at a comparable scale, ideally with a manager or director layer beneath you, and the ability to attract and keep senior talent are also required. You should be comfortable delegating hands‑on work while retaining full accountability for outcomes. Experience applying LLMs in highly regulated fields such as law or tax, and a background in model safety, adversarial robustness and red‑team policy setting, are valued but not mandatory. The role is hybrid, based in either Zug, Switzerland or London, UK, and offers a flexible work model, generous vacation, mental‑health days, retirement savings, tuition reimbursement and other industry‑competitive benefits. Thomson Reuters emphasizes continuous learning, inclusive culture and social impact, providing ample opportunities for personal growth and community contribution. What the role asks for: - Led a production LLM program end‑to‑end - Fluent in post‑training methods, distributed training and evaluation infrastructure - Proven record of shipped models, open‑source work or top‑venue publications - Led an organization of similar size with manager/director layer - Delegated hands‑on work while retaining full outcome accountability - Experience applying LLMs in law, tax or regulated domains (nice‑to‑have) - Experience setting model‑safety policy, adversarial robustness, red‑team (nice‑to‑have)