Senior Researcher (Montreal)

Senior Researcher (Montreal)

18 Nov
|
Polarized Capital
|
Montreal

18 Nov

Polarized Capital

Montreal

Polarized Capital

Tim & Dan here. Between the two of us, we’ve interviewed thousands of people and hired close to one hundred. We think normal job descriptions are dry and rarely convey what it’s actually like to work with the people hiring you.

We love building together, challenge each other constantly, and have an enormous amount of respect for one another. We’re looking for someone who brings that same combination of curiosity, conviction, and willingness to be wrong.

In this role, you’ll report to Dan and work closely with Tim and our research and engineering teams. You’ll help decide which questions are worth pursuing, design the experiments to answer them, and build the systems needed to find out.

At a glance

Title: Senior Researcher

Initial focus: World models, representation learning, and agent behavior

Reports to: Dan, Managing Director of Polarized Capital

Location: Montreal; currently remote, with a future hybrid schedule

Compensation: $150K–$200K CAD, depending on experience

About Polarized Capital

Polarized Capital started as a way for Tim to invest in, build, and learn alongside founders tackling problems that mattered most to him. Today, our interests span the human experience, science and biology, and AI and technology.

Our approach is to investigate from first principles, develop a point of view, and build something with what we learn.

We’re developing a nonprofit research lab with a longer-term ambition to advance human connection through research at the intersection of AI, human relationships, and therapeutics. Our initial AI research explores how machines learn from experience, represent systems, and predict the consequences of actions.

You’ll help turn that ambition into a credible research program through focused questions, working experiments, and results other people can examine.

About the role

This is an early research hire with meaningful ownership over both the questions we pursue and how we pursue them.

Our starting questions include

- Can a model learn from previous agent experiences to predict whether an agent will succeed on a recent task?
- Can learned representations predict how changes to software or an environment affect agent behavior, success rates, token consumption, and execution time?
- How does predictive performance change as we increase data, model capacity, or compute?
- What transfers across tasks, agents, and environments?
- When does a learned world model provide enough value to justify its complexity?

These are research questions. You’ll be expected to refine them, challenge their framing, and propose better ones as the evidence develops.

You’ll work across literature review, experimental design, data analysis, model implementation, and training. We need someone who enjoys the full process, including the unglamorous work of debugging a dataset or discovering that an evaluation doesn’t measure what we thought it did.

What you’ll do

- Turn broad ideas into precise hypotheses, learning objectives, and manageable experiments.
- Read relevant research critically, reproduce promising methods, and explain what we should learn from them.
- Build and train models using modern machine-learning frameworks.
- Develop datasets from agent traces, task outcomes, and controlled experiments, working with engineers to improve instrumentation and data quality.
- Establish strong baselines before investing in more complex approaches.
- Design evaluations that distinguish useful signal from run-to-run variability, judge inconsistency, data leakage,



and other sources of misleading results.
- Study how performance changes with data, model size, and compute, including where apparent scaling trends break down.
- Test generalization on held-out tasks, agents, and environments.
- Make thoughtful decisions about experiment cost, sample size, and what to test next.
- Keep code, configurations, datasets, and findings organized so others can reproduce and build on your work.
- Share early drafts and results, explain uncertainty clearly, and recommend whether to continue, change direction, or stop.
- Work with engineering teams to turn useful findings into practical capabilities.

What you bring

Must haves

- Experience carrying a machine-learning research project from an open question through implementation, experimentation, and a clear account of the results.
- Strong foundations in machine learning, probability, statistics, and experimental design.
- Strong Python skills and hands-on experience with PyTorch, JAX, or an equivalent framework.
- Experience training, evaluating, and debugging models, including understanding why an experiment failed.
- The ability to read a paper, interrogate its claims, and implement the parts that matter.
- Good judgment about baselines, metrics, dataset construction, and the strength of evidence required to support a conclusion.
- The independence to make progress with incomplete information and the communication skills to bring others along.
- Intellectual honesty, discretion, and care when working with sensitive data.

A PhD or research-focused master’s degree in a relevant field is valuable. Equivalent evidence of research ability matters just as much. We’ll want to understand what you personally contributed, the decisions you made, and what you learned.

Nice to haves

- Experience with world models, self-supervised learning, representation learning, or predictive latent architectures.
- Experience with reinforcement learning, planning, agent learning, or learning from execution traces.
- Experience studying scaling laws, sample efficiency, uncertainty, or generalization.
- Experience building evaluation environments, simulations, or research data pipelines.
- Publications, open-source implementations, or independent work that demonstrate original thinking and careful experimentation.

We expect the depth to investigate a difficult problem and the range to learn what it requires.

Day-to-day

Some days you’ll be reading papers and questioning whether we’ve framed the problem correctly. Others you’ll be writing training code, inspecting agent traces, running experiments, or investigating an unexpected result.

You’ll have focused research time alongside regular conversations with Tim and the team. We expect you to bring early thinking into those conversations, ask for feedback, and leave with a clearer next experiment.

We value momentum, but useful research takes patience. Your job is to choose experiments that reduce uncertainty and help us decide where deeper investment is justified.

What success looks like

Early on, you’ll help us establish a clear research question, understand the available data,



build a reproducible experimental setup, and produce a credible baseline.

Over time, success means:

- We can explain what we’ve learned, how confident we are, and what remains unresolved.
- Our experiments support the claims we make.
- Promising ideas become stronger through replication, comparison, and testing in new settings.
- We recognize weak directions early and redirect our effort.
- Your work gives the team a better foundation for future research and engineering.

A carefully established negative result can be valuable. We care about whether it changes our understanding and improves the next decision.

You’ll like this role if…

- You’re energized by a difficult question without an obvious recipe.
- You enjoy both understanding an idea and implementing it.
- You can become deeply invested in a hypothesis and still let it go when the evidence changes.
- You have a point of view and welcome people who challenge it.
- You use AI extensively while taking responsibility for checking its work.
- You care about clear writing and reproducible experiments.
- You’re comfortable helping shape an early research environment.
- You care deeply about excellence and the people you work with.

What to expect from us

From Tim: Expect curiosity, ambitious ideas, and a willingness to question assumptions. I often speak first and think second; that’s how I form ideas and stress-test my thinking. I may sound confident while still developing a view. Tell me when I’m wrong, and show me why. I’ll invest in your growth and expect you to take ownership of it. Above all, expect me to lead with integrity.

From Dan: Expect preparation, organization, direct feedback, and a strong bias toward action. Bring a point of view, share your work early, and make the next step clear. When something goes wrong, own it, learn from it, and help fix it. I’ll create room for you to do exceptional work and trust you to exercise judgment about how it gets done.

From both of us: you should be able to question the premise of a project, report a disappointing result, and ask for help. We expect the same honesty and respect in return.

Location

This role is currently remote and requires you to be based in the Montreal area. We plan to open a Montreal office, at which point the role will likely transition to a hybrid schedule. Occasional travel may be required for team collaboration or research activities.

The process

We expect candidates to evaluate us as carefully as we evaluate them.

1. Initial conversation with Dan: Your interests, ownership, judgment, communication, and what you want from your next role.
2. Research conversation with Tim: Walk us through a project you know deeply: the question, your contribution, the experimental choices, the results, and what you would do differently.
3. Focused research exercise: Work through a realistic research question or a small dataset. We’ll evaluate your framing, experimental judgment, analysis, and communication. Use AI as you normally would and be ready to explain and defend your work.
4. Final working session: Discuss your findings and challenge assumptions together.

What we’d like to leave you with It’s an extraordinary time to pursue questions that once felt impossible. We’re building a team that wants to understand something important, do exceptional work with that understanding, and build relationships that make the process worthwhile.

If you’re excited by the opportunity to help shape a research program from the beginning, apply.

📌 Senior Researcher (Montreal)
🏢 Polarized Capital
📍 Montreal

Reply to this offer

Impress this employer describing Your skills and abilities, fill out the form below and leave Your personal touch in the presentation letter.

Subscribe to this job alert:

Get the latest job offers by email for: senior researcher (montreal) / montreal

Subscribe to this job alert:

Get the latest job offers by email for: senior researcher (montreal) / montreal