The Intelligent Search Company (TISC) is building world models for human performance: AI that understands how people move, decide, and act together in fast-moving, adversarial environments. Founded in 2024 by former Wish engineers, TISC is based in San Francisco with co-founding team members in Toronto. TISC starts from a simple belief. The most important split-second decisions, the ones that affect games, missions, and lives, are still made on hunches and rigid playbooks. TISC is building the intelligence to change that. TISC treats sports as its laboratory. A game is one of the richest multi-agent environments in the world: dozens of people moving, competing, and coordinating at once, with messy video, real stakes, and instant feedback from coaches on the sideline. TISC’s models turn that chaos into understanding. They infer what’s happening, predict what happens next, and let coaches test “what if” scenarios before they make the call. But the research isn’t about sports. At its core, TISC is working on one of the hardest open problems in AI: how physical environments with multiple agents, whether humans or robots, evolve, interact, compete, and collaborate. The same models that read a defensive scheme today are built to support emergency response, search and rescue, and defense tomorrow. The culture is research-first. When a research intern came back one morning with an idea nobody had considered, a low-cost way to make video generation models more physically realistic, the team let him chase it. Seven weeks later, the results are being published. The data is unique. The problems are unsolved. What’s needed now is the researchers to crack them. Highlights:
- Founded May 2024; HQ in San Francisco with a strong Canadian team in Toronto
- $2.1M USD pre-seed (November 2025), co-led by OVO Fund and N49P, with Panache Ventures and Think+ Ventures also backing the company.
- 30+ teams have used the platform, including teams across the NBA, NFL, WNBA, and the ACC, Big Ten, SEC, Big 12, and Pac-12 conferences
- Deployment with the Women’s Premier Basketball Association (WPBA)
- Founders bring experience from Uber, Meta, Intuit, and Wish
- Expanding beyond sports into defense, emergency response, and search and rescue
- Building its founding research team, with equity set aside for its first researchers

About the Role
This role exists to push the frontier of world models. Everything else in it serves that. As a Machine Learning Researcher , you’ll join TISC’s founding research team, reporting to Co-Founder & CEO Arpan Bhattacharya. You’ll build foundation models for tactical understanding: systems that infer what’s happening,



predict what happens next, and reason about coordinated multi-agent behaviour from video and tracking data. The work spans world models, trajectory prediction, fine-grained action and intent understanding, multimodal representation learning, and video generation for simulation, forecasting, and counterfactual analysis. TISC has a leading model for predicting player trajectories in American football, and the next step is teaching it to account for each player’s individual tendencies. You’ll own that work end to end, from what data is needed to how the training recipe evolves. This is a hands-on, individual contributor research role, not an academic one. Research stays research: the founders turn your outputs into product, so your focus is on breakthroughs. There’s a clear research mandate, with three active tracks. If you find a promising direction nobody anticipated, you’ll get room to chase it.

Location:
Remote from Canada and the US World Models & Multi-Agent Research
- Build and advance foundation models that understand, predict, and simulate coordinated behaviour between multiple agents
- Push forward trajectory prediction, starting with modelling individual player tendencies in American football
- Explore video generation and simulation that let users play scenarios forwards, backwards, and counterfactually
- Advance multimodal representation learning across video, tracking and event data Applied Research on Real-World Data
- Turn novel research into systems that work on messy, high-stakes real-world data, not just clean benchmarks
- Own the full research loop: the data you need, how the training recipe evolves, and how results are validated
- Focus on the planning and decision layer: modelling what will and won’t work within simulation Research Direction & Creative Freedom
- Drive progress within TISC’s research mandate and its three active research tracks
- Propose and pursue new directions when you see something promising
- Define entirely new research problems alongside coaches, analysts, and engineers, instead of only improving existing benchmarks Publishing & Research Community
- Publish breakthroughs at top conferences like CVPR and NeurIPS, with full company support, when the work isn’t proprietary




- Help raise TISC’s profile in the research community and attract the next wave of researchers About You: You’re a researcher who builds. You’ve worked directly on models, not just the infrastructure around them, and you have the raw horsepower and research instincts to make real progress on hard, open-ended problems. You’d rather do interesting work with a small, ambitious team than be one name among hundreds at a big lab. Raw intelligence and research ability matter more than years of experience or brand names.
- Core modelling

experience:
you’ve built and trained the models themselves, not just the data pipelines or infrastructure around them. This is the one thing we won’t compromise on.
- Domain depth: experience in world models, video models, embodied AI, generative modelling, reinforcement learning, multi-agent systems, or trajectory prediction
- Hands-on builder: you write and ship research code, and you’re fluent at using AI tools to move faster. This is an IC role, not a purely academic one.
- Research taste: you know which problems are worth solving and how to make progress on them
- Real-world mindset: you’re excited by messy, high-stakes data and by problems where the answer isn’t defined yet
- Clear communicator: you can explain your work to coaches, analysts, and engineers, not just other researchers
- Startup-ready: you’re energized by building at an early-stage, founder-led company, and you value upside and impact over a big-lab title
- Based in Canada or the US Bonus:
- PhD in ML, robotics, computer vision, or a related field
- Experience at a respected research lab (e.g., Google DeepMind, NVIDIA Research, Samsung AI, Vector Institute, Tesla Optimus)
- Publications at venues like CVPR, NeurIPS, ICML, or ICLR
- Experience with video generation, trajectory prediction, or physical plausibility in simulation
- A love of sports (genuinely optional, most of the team is more excited about the broader problem) The Extras that Matter: You’ll be one of TISC’s first researchers, shaping both the models and the research culture from day one. Beyond that:
- Competitive base salary
- Meaningful early equity, with a pool set aside for TISC’s founding research team
- Adaptable start date for the right person
- Support to publish at top conferences when the work isn’t proprietary
- Unique video and tracking data that most researchers never get access to
- A direct line to the founders, with real influence on the research agenda
- The chance to solve multi-agent world modelling with impact from sports to emergency response and defense

📌 Machine Learning Researcher (Canada)
🏢 The Leadership Agency
📍 Canada