AI/ML Engineer

We're looking for an AI/ML Engineer to join our product engineering and applied research team. Our team operates at the intersection of research and production, solving open-ended problems while delivering real-world products. We use the best tool for the job, whether that's a novel machine learning model or plain software engineering fundamentals.

You're excited to build an AI product that pushes the boundaries of what's possible in education. You believe that great products and research strengthen each other. You'll thrive here if you enjoy shipping AI features into a real product, designing the evaluations that prove they work, and turning fuzzy product questions into measurable experiments. You'll make an edit in your Jupyter notebook this week, deploy it on-device next week, and watch it bring smiles to kids' faces a day later.

What you'll do

• Help define the impact of AI on children's education.
• Own AI systems and features end-to-end: from identifying product opportunities, to building and operating research and evaluation systems, to shipping production improvements used by children worldwide.
• Collaborate with learning experts and product engineers to build infrastructure measuring the educational quality, engagement, safety, latency, and behavior of an AI teacher
• Design and improve the agent harness that powers Ello's tutoring experience, including how the tutor decides what to teach next, when to intervene, and how to adapt to each child
• Develop learner profiling and adaptation systems that build a model of each child – their strengths, gaps, pace, and engagement patterns – and use that model to drive instructional decisions
• Develop feedback loops and data flywheels that continuously improve product quality through usage, evaluation, and experimentation

Who you'll be joining

Ello's 5-person applied ML team includes some of the world's foremost experts in their areas. We work on a broad range of machine learning problems across agent reliability, learner adaptation, evaluation, and multimodal perception – including building the world's best child speech recognition system, designed by co-authors of wav2vec. We collaborate closely with advisors from Stanford and top industry research labs (on a weekly basis, not in some hands-off way). We work in-person out of our San Francisco office and we're big believers in face-to-face collaboration.