Machine Learning Engineer

Prodigal is the connected AI platform leading financial institutions use to run their operations.

We work with banks, lenders, credit unions, and other financial companies that lend money to people and manage those relationships over time.

These institutions make millions of high-stakes decisions every day. Who should they reach? When should they reach them? What should they say or offer? When should a case move to a human? How should that change based on the borrower, the account, previous interactions, and the regulations involved?

Getting those decisions right requires a deep understanding of the people, processes, rules, and edge cases behind them.

Prodigal has spent the last eight years building that understanding. More than a billion interactions between financial institutions and their customers have shaped the intelligence, guardrails, and AI agents we now run in production across North America.

Today, our AI agents analyze conversations, capture context, guide human agents, decide the next action, conduct customer conversations, orchestrate outreach, and help people complete payments and resolutions. They are connected, so what is learned in one interaction can inform what happens next.

We are expanding this swarm of AI agents across more of the work financial institutions do: originations, document processing, back-office workflows, servicing, and other critical operations where money, identity, people, and regulation intersect.

We are backed by Y Combinator, Accel, and Menlo Ventures, and work with 100+ financial institutions across North America.

The Data Science and Machine Learning (DSML) team owns the full lifecycle of Prodigal's ML systems, from data curation to model selection, fine-tuning, and post-training, through to deployment, evaluation, and ongoing monitoring. We sit close to real product and customer problems, and we're expected to move from experiment to production with both speed and rigor.

• →Understand the problem domain. Work closely with product, engineering, and customer-facing teams to turn ambiguous, real-world problems into well-scoped ML problems.
• →Curate datasets. Source, clean, and structure the data needed to train and evaluate models, often before any "off-the-shelf" dataset exists.
• →Train models across the full spectrum. Build and train everything from simple classifiers to neural networks, transformers, and small/large language models, choosing the right level of complexity for the problem at hand.
• →Deploy and scale for inference. Take models from notebook to production. optimizing for latency, cost, and reliability at scale.
• →Evaluate and close the loop. Design evaluation frameworks that actually measure what matters, and build learning loops.

• Strong fundamentals in ML/DL, backed by hands-on experience training and shipping models, not just coursework or competitions.
• High agency, you take an ambiguous problem and run with it rather than waiting for it to be fully specified.
• A first-principles approach to problem-solving – you reason from the ground up rather than defaulting to whatever's fashionable or familiar.
• Genuine curiosity and the comfort to operate at the edge of what you already know.

• Real problems, real models, real data. No toy datasets, no academic benchmarks — you'll work on problems that directly affect customers and the business.
• Frontier technology, applied. Learn to harness the latest in AI and put it to work on problems that matter.
• Smart people, fast shipping. Work alongside a sharp team, ship quickly, and grow faster than you would elsewhere.