Research Scientist​/Senior Research Scientist Fundamental Research team

Research Scientist / Senior Research Scientist on the Fundamental Research team

New York, NY (US)

Toronto

$150,000 - $250,000 + Equity + Benefits

Full-Time

Vijil is the trust infrastructure enterprises need to deploy AI agents with confidence. Our platform — Diamond, Dome, and Darwin — helps organizations evaluate agents before deployment, enforce governance at runtime, and continuously harden them using production telemetry. Founded in 2023 by senior leaders from AWS and backed by Brightmind, Gradient, and Mayfield, Vijil’s platform bridges the gap between capability and trust, enabling AI agents to move from concept to production seamlessly.

With $23M in recent funding and a Gartner Cool Vendor designation, Vijil is building the future of autonomous resilience, helping companies move from prototype to production in days rather than months.

• Drive original research. Define and pursue research projects at the intersection of trustworthy machine learning, generative modeling, and agentic AI.
• Publish and present. Publish your findings at top-tier machine learning venues (e.g., NeurIPS, ICML, ICLR, AISTATS, UAI, ACL, EMNLP, NAACL, FAccT, COLT) and present them at conferences, workshops, universities, and industry events.
• Work with external collaborators and maintain relationships with relevant research labs and key individuals as appropriate.
• Translate research into product. Collaborate with applied scientists and engineers to turn prototypes into production capabilities that ship in the Vijil platform — measurably improving the reliability, security, and safety of the AI agents our customers deploy in the real world.
• (Senior) Lead and mentor. At the senior level, you will additionally set research direction for sub-areas of the team's agenda, mentor junior researchers and applied scientists, supervise external collaborations, and represent Vijil's research in external forums.

The Fundamental Research team is led by Tim G. J. Rudner and is responsible for the long-horizon scientific agenda  team’s mandate is to develop the foundational research insights that enable verifiable trust in AI agents at scale: identifying agent-specific failure modes, improving the robustness of multi-agent systems, advancing scalable oversight of frontier AI models, and creating the technical artifacts that make effective AI governance possible.

We work in close collaboration with academic partners at the University of Toronto, the Vector Institute for Artificial Intelligence, the University of Oxford, and other leading research institutions. Our research directly informs the tools that Vijil's customers depend on, and our research is released openly through publications, technical reports, and open-source artifacts on Git Hub and Hugging Face.

You are a passionate and talented researcher with a strong foundation in machine learning and a proven ability to conduct impactful, independent research. You embrace change and thrive under ambiguity, and you have a collaborative mindset and are excited to work as part of a team to tackle ambitious research challenges. You are eager to see your research translated into systems that have real-world impact, and you are driven by a desire to ensure that frontier AI is developed in a way that is reliable, safe, and broadly beneficial.

In order to set you up for success, we look for the following:

• PhD in Computer Science, Statistics, or a related technical field — or equivalent industry research experience.
• Strong publication record at top-tier machine learning venues (e.g., NeurIPS, ICML, ICLR, AISTATS, UAI, ACL, EMNLP, NAACL, FAccT, COLT) demonstrating sustained, independent research contributions.
• Deep technical foundations in machine learning, statistics, and probability — with the mathematical maturity to read, evaluate, and contribute to the cutting edge of the field.
• Strong programming skills in Python and proficiency with at least one major deep learning framework (PyTorch, JAX, or equivalent); experience using agentic coding tools; the ability to design and execute…