Internship: Junior Developer - AI Operations

Job Description

Job Description

This is a Paid Internship Position

About Mesa Quantum

Mesa Quantum is building chip-scale quantum sensing systems for resilient timing and navigation. Our work

sits at the intersection of atomic physics, photonics, RF electronics, and precision manufacturing. Backed

by top-tier deep-tech investors and the U.S. Department of Defense, Mesa Quantum is rapidly scaling and

actively recruiting exceptional innovators who want to build the future of quantum technology from the

ground up.

The Role

Mesa Quantum's knowledge is spread across nine platforms, and a lot of it lives only in people's heads.

MQoS is our answer to that — a growing, queryable information layer over the whole company. It runs on

a governed API with row-level security and reaches into every authoritative source we have: parts and

orders, lab measurements and instruments, GitLab repositories, vendor activity, documents, conversations.

The point is that a question gets a live answer, instead of a two-day hunt or a tap on somebody's shoulder.

MQoS is the largest piece of a wider surface that one person currently builds and operates: the network and

access that let engineers reach lab instruments from anywhere; the project-management and reporting

spine the team plans in; the engineering tooling that unblocks chip design; and the AI workflows that tie it

all together. That one person is now the constraint on all of it.

We are hiring a Junior Developer for AI Operations to be the hands on the keyboard across that whole

surface. You will be handed things that are already designed — a schema, an endpoint, a server to stand

up, a workflow to automate — and be very good at implementing them well and getting them into

production. You will also be the person a physicist or an engineer goes to when something they depend on

stops working.

This is not a single-product role, and it is not a help-desk role. Roughly half the job is building; the other half

is making the people around you faster. You will use Claude Code every day — we take that seriously, we

will teach you properly, and every change is reviewed before it ships.

What You'll Do

• Extend MQoS — new domains, new sources, and the migrations, API endpoints and websurfaces that
• make them answerable
• Stand up and operate internal infrastructure: Linux servers, containers, backups, monitoring, VPN and
• network access
• Keep engineers and physicists able to reach lab instruments remotely, and keep it working as the setup
• changes
• Build and maintain the Notion project-management and reporting spine the team runs on — structured
• databases, dashboards, sprint automation — and fix it the day someone says it is broken
• Get lab measurements off local instrument machines into the database, and into dashboards people
• actually open
• Install and operate engineering and vendor tooling — license servers, integrations, the unglamorous
• plumbing that unblocks a design team
• Turn research code and one-off scripts into maintained tools, with tests, configuration, and a runbook
• Build AI-assisted workflows that do real work, plus the monitoring that proves they are actually doing it

The one thing we screen hardest for

You drive the AI. It does not drive you. Claude will produce a plausible first draft of almost anything here,

and it will do it fast. What it will not do is care whether the answer is true. The people who are good at this

direct the work, interrogate what comes back, and test it before they believe it. The people who aren't

produce confident, wrong results faster than anyone can catch them.

That leads to the habit we actually hire for: building things that tell the truth about themselves. A

database replica here sat 73 days out of date behind a green health check. A service that reads vendor

email stopped reading it and went on reporting itself healthy. Software that is broken and says so is a small

problem. Software that is broken and reports success is how a company makes decisions on numbers that

aren't true. We want work that fails loudly, explains its own reasoning, and can be verified by someone who

wasn't there when you built it.

Expect us to ask about something you built or were handed, what you checked before you trusted it, and

how you would know if it quietly stopped working.

Requirements

Qualifications

• Able to follow multi-step logic, work out where a process goes wrong, and explain why — this matters more than any language you already know
• Comfortable being handed unfamiliar territory: a database one day, a network setting the next, a vendor's installer the day after
• Plans before building and finishes what they start: spec, branch, tests, review, staging, production
• Can read a written specification and implement it faithfully, and asks about what's underspecified instead of guessing
• Hands-on coding experience in any language, and able to read code you didn't write
• Willing to learn SQL properly, and careful enough not to run anything against production you haven't tried elsewhere first
• When you hit something you have never done — a Linux service, a vendor installer, an unfamiliar database — you work through it with Claude and come out understanding what you did, not just that it worked
• Treats a colleague's broken dashboard as a real problem — half this job is unblocking other people
• Comfortable saying "I don't understand that" out loud to a PhD physicist
• Helpful but not required: TypeScript, Python, Postgres, Docker, CI, Git branching, networking, time-series data, or any experience working near hardware
• No physics background, CS degree, or prior industry experience required — our last lab intern was an undergrad who used Claude to create a software tool that the photonics lab now relies on for a core capability, and which accelerated development 8x

This is an internship that pays $20–$25 per hour, based on relevant technical and/or industry experience.

Mesa Quantum is an equal opportunity employer and we welcome applications from all backgrounds regardless of race, color, religion, sex, national origin, ancestry, age, marital status, sexual orientation, gender identity, veteran status, disability, or any other classification protected by law.