Seventy-two roles across eleven families. Each says what the work is for, what evidence of capability looks like, and the stage that has to arrive before it starts. Most are not open yet, and each one tells you plainly which it is.
A catalog is not a hiring plan. Of the 72 roles below, only the ones marked Open now are funded and receiving applications today; those live on the openings page. Everything else is work we intend to do, held behind a stage that has not arrived yet, and each entry names the stage. We would rather show you the whole map and tell you where we actually are on it than publish seventy-two positions we are not filling.
Every role is described by what it is for and what evidence of capability looks like, not by years of experience or a degree. If the evidence describes you, that is the conversation worth having, whatever your background says.
All 72 roles in the catalog.
Help a person own a computer they can trust with their working life. You will make the hardware, operating system, private data, agent and remote execution behave as…
Turn the ambition of personal intelligence into questions that can be answered honestly. You will connect foundational research to useful capabilities people can und…
Make powerful personal computing practical enough to live in a home. You will own the complete device experience, from component choices to noise, recovery and repai…
Earn the right to hold a person's most sensitive information. You will make protection, consent and recovery requirements part of every engineering decision.
Make the user's computer dependable when the agent is busy and the person needs it most. You will work close to the operating system.
Let people add useful tools without giving every tool access to their lives. You will build the isolation around models, plugins and remote jobs.
Help one person's agent work across many machines without losing track of who is in charge. You will make delegation and recovery predictable.
Give people access to computers they own or are authorized to use, wherever those computers are. You will make remote work secure and observable.
Keep a person's information usable and recoverable for years. You will build the storage foundations beneath their memory and files.
Keep personal agents useful after the first demonstration. You will make a fleet of household computers maintainable without treating users' private data as telemetr…
Make every delivered component traceable to an intended source and build. You will protect the path from an engineer's change to a user's computer.
Discover which hardware choices improve personal intelligence. You will connect real workloads to memory systems, accelerators and the software that uses them.
Turn a proven architectural idea into hardware that computes correctly. You will design the digital logic for a justified accelerator or subsystem.
Find hardware mistakes before they become permanent. You will make the evidence for correctness as important as the architecture itself.
Make an approved chip design manufacturable and reliable at its intended operating limits. You will connect logical intent to physical implementation.
Make sure every shipped chip behaves correctly beyond the simulator. You will connect design-for-test, bring-up and production learning.
Test hardware ideas early enough to change them. You will build prototypes that let software and architecture teams learn before committing to custom silicon.
Move information efficiently between the parts of a computer. You will work on the memory and packaging choices that determine practical model capacity and performan…
Make the physical signals beneath digital computing stable and trustworthy. You will evaluate or design the analog and power circuits justified by the product.
Investigate whether emerging interconnect technologies can improve future personal or connected compute systems under realistic cost and power limits.
Build computers that behave as carefully in a home as they do on a laboratory bench. You will turn system requirements into working electronics.
Make the computer start safely, understand its devices and recover when an update fails. You will work where hardware and software meet.
Make serious computing comfortable to live with. You will design for sustained performance, safe temperatures, quiet operation and physical durability.
Make personal intelligence affordable to run throughout the day. You will connect useful work to energy use and household operating constraints.
Keep personal devices connected in ordinary homes. You will make wireless performance reliable when walls, interference and device placement complicate the design.
Make the hundredth computer as trustworthy as the first. You will translate engineering intent into repeatable assembly, testing and quality.
Keep users' computers available, supportable and repairable over time. You will build the supply and service relationships behind the product.
Make useful intelligence run responsively on a computer the user owns. You will turn models into a dependable local service.
Make each unit of compute do more useful work. You will optimize the kernels and compilation paths that matter to the user's real workloads.
Let users adapt models and run appropriately sized training jobs on resources they control. You will make training reproducible, interruptible and accountable.
Make research results and model releases traceable to the data and decisions that produced them. You will build the infrastructure for learning responsibly.
Make performance claims useful to the person buying the computer. You will measure what the system accomplishes, what it costs and how reliably it works.
Explore how much useful intelligence can fit into an affordable personal computer. You will pursue improvements that survive honest evaluation.
Help a private agent learn what matters to its user while preserving control and the ability to change their mind.
Help an agent finish meaningful work while recognizing when it needs a person. You will study planning under uncertainty and limited authority.
Help independent agents cooperate without losing the intentions or boundaries of the people they represent.
Make personal computing accessible through the ways people naturally communicate. You will study speech, vision and other inputs under real device constraints.
Find ways to learn from useful signals without casually exposing the people behind them. You will test privacy methods against explicit adversaries.
Show when a private agent deserves a user's trust. You will build evidence about task completion, failure and human oversight.
Build the part of Private Agent One that turns a person's intention into completed, inspectable work.
Help a person find and use their own information without losing its source, meaning or privacy.
Make 'my agent' mean something precise. You will define who can ask for an action, who may perform it and how permission ends.
Protect personal information without making a lost device the end of a person's digital life. You will design key custody and recovery.
Give the private agent a trustworthy foundation beneath the operating system. You will connect device identity and boot integrity to usable security decisions.
Keep documents, websites and tools from silently taking control of a person's agent. You will secure the boundaries around useful automation.
Find the failures that matter before users encounter them. You will test the product as a connected hardware, software and human system.
Make the most important system guarantees precise enough to check. You will focus formal reasoning on boundaries where mistakes are costly.
Make ownership visible in everyday operations: access, correction, export, sharing and deletion. You will turn those promises into working systems.
Help Private Agent One work with other agents and tools while preserving the user's authority. You will build the connections and test their behavior.
Connect a person's agent to the services that already run their life and business. You will make ordinary workflows dependable across imperfect interfaces.
Make personal agents useful inside organizations that have their own security and operating requirements. You will bridge real deployments and the product team.
Help other builders extend personal computing without needing to understand every internal subsystem. You will make the safe path easy to use.
Earn a community by helping people build useful things. You will connect technical contribution, documentation and respectful collaboration.
Choose the problems worth solving first. You will turn the broad mission into a product people can understand, use and afford.
Make powerful computing feel understandable and calm. You will design how a person sets an intention, grants permission and sees what happened.
Understand how people and their agents can cooperate without weakening human judgment or personal boundaries.
Make private computing feel at home on the Apple devices people already use. You will build polished, secure native experiences and integrations.
Make the same personal-agent promise work across diverse computers. You will handle the platform details that determine reliability and trust.
Help a person's agent travel with them across phones and small devices. You will connect local capabilities to the wider personal system.
Give people a clear window into the work their private agent is doing. You will build the interfaces and services they use every day.
Help the product succeed in real homes and workplaces. You will turn installation and support into a source of engineering understanding.
Bring the best available components and engineering partners into one coherent personal-computing product.
Help users extend their own computing through partners without losing choice. You will align external capabilities with the product's trust model.
Build lasting relationships between researchers and a product with important unanswered questions. You will make collaboration worth the university's time.
Find people whose work and values can advance the mission, and give them a thoughtful experience from first contact to decision.
Take responsibility for a difficult technical boundary that spans teams. You will earn influence through judgment, evidence and the people you help grow.
Bring a rigorous academic perspective to a real personal-computing problem while retaining a clearly defined university role.
Spend a focused period answering a hard question alongside the engineers building the system. You will have a specific research home and deliverable.
Turn promising ideas into experiments that other people can trust and extend. You will work between scientific exploration and production engineering.
Use your research depth on a question that matters to a real user. You will join a scoped project with a named mentor and realistic resources.
Learn by building something useful with experienced engineers. You will own a bounded contribution with close mentorship.
If an important part of this mission is missing from our role list, help us see it. This is a route to discuss a contribution and identify a suitable funded role.
Technical leadership · Systems software · Silicon and architecture · Physical computers · AI infrastructure · AI research · Private information and trust · Connected systems · Human experience · Partnerships and talent · Research and contribution routes
This catalog is reviewed through 2026-12-09. Roles are added and retired as the work changes; identifiers are stable, so a role you were sent keeps its address.