Scalrr × Vijil / Candidate briefing
Make agents
worthy of trust.
Lead the research.
Head of Applied Research
Turn ambitious ideas into trustworthy AI systems. Build the team, define the experiments, and bring the work into the world.
01 / The company
Trust is the next frontier.
Vijil helps enterprises build and operate AI agents they can rely on in mission-critical work.
Founded by leaders behind Amazon SageMaker’s deep learning platform, Vijil connects AI research with the practical demands of enterprise deployment: reliability, security, safety and governance.
A consequential problem
As agents gain tools, autonomy and the ability to work together, evaluating a single model is no longer enough. Enterprises need to understand how whole systems behave—and keep them within acceptable boundaries.
An early leadership opportunity
Set the applied research direction, build the team and work directly with the CEO. Shape both the scientific methods and the capabilities that reach customers.
Backed for the next chapter
Investors include Mayfield, Gradient and Brightmind. Funding includes a $17M Series A following a $6M seed round.
The question at the center of the role
How can agents and multi-agent systems improve their capabilities without losing reliability, security or control?
02 / The platform
From evaluation to evolution.
A modular, API-first approach to understanding, protecting and improving enterprise agents.
Discover · Identify
Inventory agents and establish their identities across an enterprise environment.
Diamond · Verify
Evaluate behavior with customized tests and adversarial probes. Build evidence for reliability, security and governance decisions.
Dome · Defend
Apply runtime controls to model and tool interactions, including what an agent can access and disclose.
Darwin · Evolve
In development with design partners: investigate agent improvement within defined limits, testing proposed changes before human approval.
Where applied research connects
Your team turns promising methods into validated prototypes and works with engineering to integrate them into the platform. Experiments must show measurable gains against strong baselines and meaningful customer problems.
03 / The role
Head of Applied Research
Own the bridge between fundamental research and capabilities enterprises can deploy.
Work as a peer to Chief Scientist Tim G. J. Rudner. Define the applied agenda, remain hands-on in experiments and prototypes, and build a team of applied scientists and machine learning engineers. VP-level scope and title may be considered based on experience.
What you’ll do
- Own the applied research agenda: Set priorities across automated red teaming and blue teaming, agent self-improvement, scalable oversight, and trustworthy AI. Determine which problems the team tackles and why.
- Develop and validate new methods: Design methods, run experiments, and demonstrate measurable improvements against the strongest existing alternatives. Establish a rigorous standard for when research is considered validated.
- Turn research into product: Partner closely with product and engineering to translate validated methods into platform capabilities. Define acceptance criteria, support engineering handoffs, and measure results with real customers.
- Run a fast research loop: Lead short, focused prototyping and experimentation cycles. Leverage AI coding tools and automated agentic workflows to dramatically reduce the time between an idea and a measurable result without sacrificing research rigor.
- Build a world-class applied research team: Hire, mentor, and retain applied scientists and machine learning engineers across the US and Canada. Establish the practices and culture that allow a distributed research organization to move quickly and operate effectively.
- Partner across research and leadership: Work closely with the Chief Scientist and fundamental research team to identify promising approaches and validate methods, while partnering with the CEO and broader leadership team to help shape Vijil's technical direction.
- Stay at the frontier: Track advances across AI agents, evaluations, guardrails, adversarial robustness, AI safety, and related fields. Determine which emerging techniques Vijil should adopt, extend, or replace.
Who you are
- Experienced research leader: 10+ years of research experience, including 5+ years leading research teams in industry or academia.
- Accomplished researcher: PhD in Computer Science or a closely related field, with a record of AI/ML research published at premier venues such as NeurIPS, ICLR, or ICML.
- Deeply technical: You can critically evaluate a research paper, scope an ambitious research project, debug a model or agent, and read, write, and evaluate Python code.
- Fluent in modern AI systems: You have hands-on knowledge of machine learning systems, AI agents, evaluations, guardrails, AI safety, and the failure modes unique to multi-agent systems.
- Research-to-product oriented: You have worked closely with engineering teams to turn research methods and prototypes into shipped products or production systems.
- Modern in how you work: You use AI coding tools and automated agent workflows regularly and understand how they can accelerate research and development.
- Effective distributed leader: You have successfully built or managed remote or geographically distributed technical teams.
Nice to have
- Experience building or evaluating agent systems at an AI-native company or research lab.
- Published or shipped work in red teaming, adversarial robustness, recursive self-improvement, evaluations, scalable oversight, or AI safety.
- A track record of research or technology that has been adopted and used by others.
- Experience taking an idea from paper or prototype through production and ultimately into something customers pay for.
- Experience leading a small, highly technical team in an early-stage startup environment.
- Strong opinions about how AI coding tools and agentic workflows should change the way research teams operate.
- Experience making high-quality technical decisions with incomplete information and introducing only the processes necessary to keep a team moving quickly.
04 / Research agenda
Ambitious questions. Working code.
Build an experimental program that connects scientific rigor with useful, measurable results.
Multi-agent systems
Study coordination, composition and emergent behavior as agent populations grow. Test what changes when tasks, environments or constraints change.
Self-improvement with control
Explore how agents can learn and adapt while remaining bounded by security and governance requirements. Define experiments and success criteria for evolutionary approaches.
Automated red & blue teaming
Develop adversarial methods and defenses. Validate robustness against strong alternatives, new attacks and realistic failure modes.
Scalable oversight
Create evaluation and oversight methods that can keep pace with increasingly capable systems. Make behavior and failure modes legible to the people accountable for them.
How the work moves forward
- Define the research question and a meaningful baseline.
- Build prototypes and experimental environments; run experiments at scale.
- Measure gains, investigate failure modes and challenge the result.
- Translate validated methods into platform capabilities with engineering and product.
- Publish research while protecting relevant intellectual property.
A hands-on research leader
This role calls for someone who can design experiments, write and evaluate Python, debug agents, and work alongside researchers. Relevant experience includes multi-agent reinforcement learning, RLVR, post-training, agent evaluation and scalable oversight; depth matters more than familiarity with every framework.
05 / The people
Build alongside experienced leaders.
Direct access to company leadership, with a close partnership between applied research and fundamental science.
Vin Sharma
Founder & CEO
Your manager. Previously GM and Director of Engineering at AWS AI, including work on Amazon SageMaker.
Tim G. J. Rudner
Chief Scientist
Your scientific peer and partner. Assistant Professor at the University of Toronto, CIFAR AI Chair and Vector Institute faculty member.
Radina Mihaleva
Head of Business Development
Previously COO of Astronomer, with experience scaling cybersecurity businesses.
Doug Rendler
Head of Sales
Enterprise sales leadership across cybersecurity and SaaS.
Steve Coplan
Head of Marketing
Experience bringing emerging security and technology companies to market.
Build the applied research organization
The planned team is approximately six applied researchers and ML engineers over the next 12 months. Establish the hiring bar, develop researchers and create a culture of autonomy, technical ownership and individual accountability.
06 / Culture & offer
Freedom to explore. Ownership to deliver.
A distributed research environment built around technical rigor, candid collaboration and customer impact.
High autonomy
Own meaningful research decisions and take responsibility for their results. Work with a small team where individual contributions matter.
Research that ships
Connect papers, experiments and prototypes to real product value. Stay close to engineering and the enterprise problems behind the work.
Room to publish
The research ambition includes approximately two to three high-quality papers per year, with publication coordinated around relevant patent filings.
Remote collaboration
Open to candidates across the US and Canada, with a preference for the San Francisco Bay Area. Build effective practices for a distributed technical team.
Compensation
Target base salary is $300,000–$350,000 and 1–2% equity. Discuss level, location, equity terms and the complete package with your Scalrr recruiter; these are search targets, not a guaranteed offer.
How Vijil works
- Start with customer problems and work backwards.
- Take ownership and use constraints to inspire invention.
- Learn relentlessly, dive into the details and benchmark against the best.
- Communicate candidly, respect colleagues and invest in their growth.
- Measure success by sustained customer impact.
07 / Interview process
Start with a conversation.
Scalrr will guide you through the search and coordinate next steps with Vijil.
Connect with Scalrr
Discuss your research background, leadership experience, interests and questions about the opportunity.
Meet the Vijil team
The process includes conversations with three interviewers. Your recruiter will confirm the participants, sequence, format and timing.
Explore technical depth
Expect a strong emphasis on your ability to do the work: define research questions, design experiments, build prototypes and translate results into practical capabilities.
Research leadership, grounded in practice
Be ready to discuss the work you personally contributed, how you evaluated alternatives, and what happened when the research met real systems. Your recruiter will share any specific preparation requested by the team.
08 / Explore more
Go deeper into the work.
Learn about Vijil’s platform and research before your conversations. For this opportunity, connect directly with Scalrr.
Vijil website
Explore the company’s approach to trustworthy agents.
Explore ↗Vijil Labs
Read about the research program and published work.
Explore ↗Developer documentation
Explore the developer-facing platform documentation.
Explore ↗Interested in the opportunity?
Contact your Scalrr recruiter to discuss the role and next steps.
Connect with Scalrr ↗