Open Role
Stealthmode - AI CTO
CO-FOUNDER & CTO (Research → Product → Deployment → Portfolio Outcomes) Stealth-stage AI consolidation company · Partnering with a two-time exit founder · Montréal / Mila-adjacent · Co-Founder, full-time
THE BUSINESS: AI-POWERED CONSOLIDATION
Canada builds some of the best AI research on earth and adopts almost none of it. The breakthroughs leave on a plane. The companies that actually run the economy — the ones employing millions of people in services, staffing, operations, finance — are still waiting for AI that does real work inside their real workflows, not a demo. We're not building one product and selling it. We're building an AI-powered consolidation. We raise capital to acquire companies, then transform them with AI — taking the manual hours out of how they operate, multiplying the output of the people already there, and compounding the economics of the whole portfolio. The AI is the thesis: it's why the acquired businesses are worth more under us than they were the day before, and why the portfolio gets stronger with every company we add. The founder is a two-time exit founder who has built, scaled, and sold companies, and knows how to take an idea from zero to deployed, trusted, and acquired. He's looking for a co-founder and CTO to own the entire AI and technology engine — research, product, deployment, and, most importantly, the outcomes it produces across every company we acquire. Our thesis is "AI for all": AI shouldn't be a luxury good for a handful of frontier labs and Fortune 50 budgets. It should be in the hands of the operators who've never had access to it — and it should make the people already doing the work dramatically better, not redundant. We take the manual hour — the search-and-stare, the copy-paste, the admin — so experts can spend their time on judgment, relationships, and the calls only a human can make. AI replaces the work, not the worker. If you want to build AI that puts people out of a job, we're not the team.
WHAT YOU'D ACTUALLY OWN
This is not a research seat. As CTO and co-founder you own the whole arc, end to end, with a number attached: • Research — set the scientific agenda for the core AI engine: ranking, retrieval, LLM systems, outcome modeling. The hard problems below are yours. • Product — turn that research into product that real operators use every day, inside real workflows, without an ML PhD to babysit it. • Deployment — get it live and trusted inside each acquired company, integrated with their systems and their people. • Portfolio outcomes — the one that matters most: prove the AI moved the P&L. Faster, more, better, cheaper — measured in each portfolio company's real results, not benchmark scores. Then make it repeatable, so company #2, #5, #10 transform faster than #1 did. Every acquisition is a new deployment of the same engine and a new chance to compound it. You own making that engine sharper, more general, and more valuable with every company we bring in — and the technical org that scales it.
WHY TECH SERVICES IS WHERE WE START
Our first vertical is technology services — IT staffing, professional services, the firms that supply skilled people to the enterprises building everything else. It's one of the best candidates for AI-powered transformation we could have picked: • It's enormous and underserved. A multi-hundred-billion-dollar global industry still running on manual search, spreadsheets, and gut feel. The tooling is a decade behind the value it moves. • The core problem is fundamentally a ranking problem. "Which of these thousands of people is the right match for this specific need, right now?" is exactly what modern ML is built to do — and exactly what no one has done well. • The feedback loop is real. Every decision produces a real outcome — submitted, interviewed, hired — so you can measure whether the model is right and improve from ground truth, not vibes. • It's fragmented and acquirable. A long tail of strong, profitable, owner-operated firms — perfect for a consolidation where AI is the multiplier that makes the combined whole worth far more than the parts. Win here, and the same engine generalizes to every services industry that matches supply to demand — which is the rest of the portfolio.
THE HARD RESEARCH AT THE CORE (THE FUN PART)
We've already solved the easy half. Retrieval surfaces the right candidate into the pool ~97% of the time. The value isn't there anymore. The frontier is ranking under sparse, delayed, real-world outcome labels. The signal that separates the person who actually gets hired from ten look-alikes with identical résumés is not in the résumé. It's off-document: availability this week, intent, the read from a screening call, the state of a relationship. The open questions you'd own: • Learning to rank from sparse converter labels — real outcomes are rare and mature over weeks. How do you train a ranker that compounds as outcomes arrive without overfitting to noise? • Leak-safe temporal features — every feature computed strictly as-of decision time. Lookahead leakage is the silent killer of every "great" offline number; permutation controls and time-rewind backtests are non-negotiable. • Capturing off-document signal — pipelines that turn unstructured human interactions (calls, notes) into structured, rankable signal via LLMs, vendor-agnostic and leak-free. • Champion/challenger MLOps for a model you can never ship blind — no online weight updates, no ungated promotions. A challenger earns production through a backtest gate or it doesn't ship. • LLM-in-the-loop ranking and retrieval — Claude as planner, modern rerankers and embeddings as the recall floor, learned re-ranking on top — and the engineering to redeploy all of it cleanly into each new company we acquire. Live production outcomes, real data, real businesses. Your work changes real decisions for real people and shows up in the portfolio's results.
THE KIND OF CO-FOUNDER WE'RE LOOKING FOR
We don't care about the length of your publication list. We care that you can take an AI idea all the way to a P&L outcome — research it, build it, ship it into a real business, and prove it worked — and that you're honest about when it didn't. You're probably a strong fit if: • You have deep applied experience in learning-to-rank, recommender systems, information retrieval, or outcome modeling — ideally with LLMs in production (RAG, agents, rerankers, embeddings). • You're a builder and a shipper, not just a modeler — you can own the research, the pipeline, the product, and the deployment, and hire and lead the technical team that scales it across companies. • You're rigorous about evaluation to the point of paranoia — leak guards, holdout discipline, "is this just noise?" controls before you celebrate a lift. • You think in outcomes, not artifacts — a deployed system that moved a real number beats an elegant model that didn't. • You want to be a founder — the 0-to-1 fog, the ownership, the upside, the risk. Acquiring and transforming a portfolio of real companies excites you more than a tidy research roadmap. • You're intellectually honest — you'll tell us a clever idea is null, kill it, and move on. Five null experiments that close a question beat one oversold win. • Bonus: ties to the Montréal / Mila community; reading-knowledge of French; genuine conviction about AI adoption in Canada.
WHAT YOU GET
• A true co-founder & CTO seat — founder-level equity in the platform that owns the whole portfolio, a peer voice on direction, and ownership of the entire technology and AI function from research to outcomes. • A partner who has done the exit twice — and a capital-backed strategy to acquire real companies, so your work deploys into real P&Ls from day one instead of chasing a single product-market fit. • A mission with a conscience: putting frontier AI in the hands of operators who've never had it, in a way that makes people better at their jobs instead of replacing them. We're early and we're picky — this is a marriage, not a hire. If owning the AI and technology that transforms a whole portfolio of companies sounds like the thing you've been waiting to do, let's talk. email: bobbi.bidochka@mila.quebec
Apply
Send a short note explaining why this venture, and link your LinkedIn or résumé.