AI Software Engineer - Calendar Intelligence

Apple has changed how people live, connect, and work, and the Calendar team wants to rethink what a calendar can do for the people using it. A calendar knows when someone is busy. It knows very little about which of those commitments matter, which ones could move, or when would be a good time for something new. We're looking for an engineer to help answer those questions, on device and privately, for billions of people. The right person has strong machine learning foundations and is just as comfortable writing the software that ships a model as building the model itself. You'll help take ideas from a single sentence to a shipped feature, and you'll bring the evaluation that tells us whether it works.

Description

In this role you'll drive the modeling and evaluation behind intelligence features in Calendar, from the first framing through launch, as part of a very collaborative team. A request will often start as one line, and you'll work with the team to turn it into a problem we can solve: what signal exists, what a good answer looks like, which approach fits, and how we'll know it's right. You'll design the evaluation before you build the model. That means the metrics, the eval sets, the baselines a model has to beat, and human review for the places where a metric can't tell you enough.

You won't always have good data, so part of the job is finding it and getting something useful out of it when there isn't much. You'll need to get real statistical strength out of priors, shared structure, and self-supervision. You'll also work with a range of large language models, on device and on server, deciding where a model call earns its cost, where a smaller model trained on a larger model's judgments does better, and where a well-tuned heuristic beats both.

You'll join a small team of senior systems engineers who ship this work across Apple's platforms, several of them with deep backgrounds in ML data engineering. There's no hard line between roles, so you'll write production code alongside them in whatever part of the stack the problem needs. Calendar also builds on signals and models from other ML teams across Apple, and a lot of this role is working with them: building on what they have, agreeing on shared interfaces, and recognizing when we need our own. You'll work closely with design and product as well. We need someone who writes clearly, can explain a model's tradeoffs to engineers who aren't ML specialists, and is comfortable driving a problem forward before the path is defined.

Minimum Qualifications

Bachelor's degree or foreign equivalent in Computer Science, EE, Applied Mathematics, or related field with 2+ years experience in developing ML / AI driven software.

Proficiency in one or more of the following programming languages: Python, Swift, Objective-C, or C++

Experience developing software prototypes and features driven by Large Language Models and designing tool and function-calling interfaces, constructing context, and validating model output

Fluency with agentic coding tools such as Claude Code, with good judgment about where to apply them and how to verify what they produce.

Experience driving an ambiguous problem with cross-functional partners, including other ML teams, from a loose idea through launch, and getting alignment on shared interfaces along the way.

Exceptional problem-solving skills, and the ability to drive alignment across teams on shared API surfaces

Preferred Qualifications

Experience designing and developing systems that are efficient in resource constrained environments - latency, memory, and power budgets on device

Solid foundation in machine learning, probability, and statistics, with hands-on experience training and evaluating models in a framework such as PyTorch or JAX

A track record of designing the evaluation for what you ship: metrics, held-out sets, baselines, and noticing when a metric has stopped measuring what you care about

Proven ability to excel in a fast-paced development team