
Lead/Senior Research Scientist, Agentic RL
Posted 1 day ago

Posted 1 day ago
This is a fully remote position, open to applicants in United States, +5 more locations.
• Develop benchmarks, training datasets, rewards, and training signals to enhance the Thomson model family.
• Create rigorous experiments and diagnostics, including ablation studies, to separate algorithmic effects from data, reward, and system artifacts.
• Translate research findings into production models.
• Construct and expand the evaluation suite to assess the strengths and weaknesses of the Thomson models.
• Utilize evaluation outcomes to guide training objectives and technical trade-offs.
• Collaborate with Research Engineers on large-scale training and evaluation infrastructure.
• Partner with subject-matter experts in law, tax, and other professional fields.
• Coordinate with product teams and model users to comprehend adoption requirements and facilitate integration.
• Publish consistently in leading ML and NLP conferences.
• Represent Thomson Reuters Labs within the research community.
• At the Lead level, manage a research agenda, convert emerging methods into a roadmap for models, and mentor scientists and engineers.
• PhD in machine learning, computer science, or a related field, or equivalent practical experience.
• Practical experience with LLM post-training and/or evaluation, ideally at the forefront of model development.
• Proficient in Python and PyTorch.
• Proven track record of research impact, including publications in top-tier venues and/or contributions that influenced an LLM development cycle.
• Experience collaborating with researchers and cross-functional stakeholders, including subject-matter experts, to translate concepts into practical training or evaluation tools.
• Exceptional written and verbal communication skills.
• Ability to navigate ambiguity in a fast-paced work environment.
• For Lead level: 3+ years of research experience post-PhD.
• For Lead level: demonstrated history of leading a research agenda to published or deployed outcomes.
• For Lead level: experience in mentoring or supervising researchers.
• Preferred: familiarity with RL post-training frameworks like NeMo-RL, verl, OpenRLHF, or TRL.
• Preferred: knowledge of evaluation frameworks such as Inspect or lm-evaluation-harness.
• Preferred: experience designing LLM-as-a-judge or rubric-based reward and evaluation pipelines.
• Preferred: familiarity with Mixture-of-Experts or long-context models.
• Preferred: experience in law, tax, or other professional non-verifiable domains.
• Preferred: skills in experimental design and working with human data.
• Preferred: knowledge of AI-assisted coding and coding agents at scale.
• Flexible work arrangements, including the option to work from anywhere for up to 8 weeks a year.
• Grow My Way programming and skills-first development opportunities.
• Flexible vacation policy.
• Two company-wide Mental Health Days off.
• Access to the Headspace app.
• Retirement savings plans.
• Tuition reimbursement.
• Employee incentive programs.
• Resources for mental, physical, and financial wellbeing.
• Two paid volunteer days off each year.
• Pro-bono consulting projects and ESG initiatives.
• Comprehensive US health, dental, vision, disability, and life insurance programs.
• Competitive 401k plan with company matching.
• Generous vacation, sick leave, and paid time off policies.
• Paid holidays, including two company mental health days off.
• Parental leave benefits.
• Sabbatical leave options.
• Optional hospital, accident, and sickness insurance.
• Optional life and AD&D insurance.
• Flexible Spending and Health Savings Accounts.
• Fitness reimbursement program.
• Employee Assistance Program available.
• Group Legal Identity Theft Protection benefit.
• Access to a 529 Plan.
• Commuter benefits.
• Adoption & Surrogacy Assistance available.
• Employee Stock Purchase Plan offered.
• Potential eligibility for an Annual Bonus based on enterprise and individual performance.
Boehringer Ingelheim
mpathic
Cogstate
Sequen
Get handpicked remote jobs straight to your inbox weekly.