Full-Stack Machine Learning Engineer

About Signal 1 US health systems will spend $100-130 billion on AI between 2025 and 2030, and studies show that 95% of AI initiatives fail to deliver measurable impact. Health systems have little to no insight into which AI solutions actually work and bring value, meaning that millions of dollars are wasted on tools that make no difference. Signal 1's AI Management System enables health systems to operate scalable and valuable AI programs that can ensure solutions are performing as expected and make data-driven decisions on where to invest. Purpose-built for healthcare and informed by experience managing AI at scale, Signal 1 empowers health systems to use AI to its full potential and maximize impact on patient care. We were founded by Tomi Poutanen (former Chief AI Officer at TD Bank, co-founder of the Vector Institute and Radical Ventures) and Mara Lederman (previously a professor at University of Toronto's Rotman School of Management and lead at Creative Destruction Lab), with early backing from Geoffrey Hinton. About the role Health systems are starting to build their own AI agents: agents that draft discharge summaries, reconcile medications, prepare prior authorization reviews, and schedule patients. Every one of those agents acts on real patient data inside real clinical workflows, and every one of them needs to be visible, evaluated, and defensible from the day it goes live. The Agent Control Plane is the latest extension of our flagship product, the Signal 1 AI Management System (AIMS). It gives a hospital one place to manage, evaluate, continuously improve, govern, and interact with all of its agents. The work behind that spans grounding agents in the organization's own context, evaluating agent behavior in a dynamic clinical environment, extracting actionable insights that enable continuous improvement, and integrating with a hospital's systems and people to enable effective agentic workflows. This part of the platform is a true zero-to-one build that we are co-developing with leading US health systems including NewYork -Presbyterian and Mount Sinai. We have a clear thesis and a small team with room for you to own large pieces of the product. You will work directly with the Director of Engineering, and the technical decisions you make will shape the architecture, the codebase, and the product for years. This role suits someone who has already taken ML-powered products from first commit to production users and wants to do it again with far more ownership. About the team You'll work with a team of engineers, machine learning scientists, and product people who are ambitious and passionate about their craft. We work hard and move fast to shape the way healthcare uses AI for the better. Our office is based in Toronto and we come together 1-2 days a week to collaborate, innovate, and have fun. What you'll do Own features end to end : Take a problem from ambiguous idea to production. You'll design the data model, build the pipelines and backend services, and work across the frontend when your feature calls for it. You own the whole slice and the outcome it delivers. Build the AI core of the product : Design and implement the systems that ground agents in a hospital's own context, evaluate their behavior in a dynamic environment, and turn agent activity into actionable insights that drive continuous improvement. Work with messy, real-world data : Ingest and normalize agent telemetry from many runtimes, healthcare data standards like FHIR, and clinician feedback, then turn it all into datasets the product and its evaluations can rely on. Prototype with design partners : Build demos and pilots alongside real health systems, watch them get used, and fold what you learn back into the product within days. Invent where there is no playbook : Agent observability and evaluation in healthcare is a brand-new field. Many of the problems you'll face have no established solution, and you'll be expected to create one, test it against real data, and share what you learned. Raise the bar : Agent management is a nascent practice that is developing and changing quickly. You'll keep our patterns for testing, evaluation, logging, and reliability ahead of a fast-moving state of the art , and the standards you set will spread through the whole team. You'll be a great fit if you have... Experience - 3+ years of industry experience shipping software, with a substantial portion of that time spent on applied machine learning - A track record of building production AI systems: you have taken LLM-powered or ML-powered features into production, operated them at scale, and been on the hook when they misbehaved - Depth in AI and backend engineering: strong Python for services and ML pipelines, working fluency with cloud infrastructure and CI/CD, and a working understanding of frontend development (TypeScript, React, or equivalent) with enough hands-on experience to work across the stack when a feature needs it - Deep hands-on data experience: you have built data pipelines, cleaned messy datasets, designed schemas, and understand how data quality issues surface downstream in models and metrics - Zero-to-one shipping: you have taken at least one product or major product area from an empty repo to real users - A history of novel solutions: you can point to problems that had no established approach and describe the solution you created, why it worked, and how you validated it Ways of working - A strong product mindset: you start from who the user is and what outcome they need, and you let that drive every technical decision - Comfort navigating uncharted waters: you make good decisions with incomplete information, timebox your bets, and course-correct quickly when the evidence points somewhere new - High agency and resourcefulness: you find the information you need, unblock yourself, and pull in help early when a decision is bigger than you - Clear, succinct communication: you keep engineering, product, and customers on the same page as things move fast, and you translate fluently between technical and clinical audiences - An AI-native way of working: you use AI tools daily across coding, research, and prototyping to speed up every step, and you hold the output to a high standard - A high bar for craft: you push beyond "does it work?" to "does it feel right?", in the product and in the codebase Nice to have - Experience with healthcare data (FHIR, HL7, EHR integrations) or other regulated, privacy-sensitive domains - Ownership of a product that shipped to production, with responsibility for iterating on it release after release toward product-market fit - Experience building agentic systems that operate on large-scale data in production - Experience conducting AI research, with peer-reviewed publications at conferences or in journals What we'll bring - Competitive compensation (base salary + bonus + equity) in a high-growth company - Comprehensive health benefits - 4 weeks PTO - A green-field product, direct access to design partners at major health systems, and a real say in what gets built To apply To apply for this role, please send an email to [email protected] with: - "Re: Signal 1 Full-stack Machine Learning Engineer" as the subject line - Your resume - A walkthrough of an ML or AI system you took to production: what it does, the hardest technical decision you made along the way, and why you chose to share this one. Write it as plain text in the body of the email, 500 characters maximum. You may attach one image to supplement the walkthrough.