Work alongside the people writing the evals.
Lattice/AI runs a small permanent team and brings in collaborators for specific work. Two of those pathways are open to students at top universities, undergraduates and graduates alike: an independent research fellowship and a short-term internship.
Both pathways are designed around publishable artifacts. Whatever you contribute carries your name on the signed report, the public scorecard, the paper, or the patent filing. Your work is not buried in someone else’s deck.
These pathways are unpaid today. We are a small firm that just launched, and we do not have the cash to pay competitive salaries yet. Read the “honest part” section below before you decide whether this fits your situation.
Independent Research Fellowship
You propose a research question, or pick one from our open list. We provide methodology guidance, dataset access where applicable, and weekly mentorship from the founder. The output is a signed research paper or eval report published under joint byline.
- Arabic and MENA-region LLM evaluation
- Agent safety and red-teaming methodology
- RAG grounding and citation traceability
- Governance framework mapping for EU AI Act, NIST AI RMF, or ISO 42001
- Open evaluation tooling
- A published artifact under joint byline
- Sponsored conference submission or journal article processing charges, paid by the firm
- Where the work yields a patentable contribution, the patent filing is sponsored and you are named as an inventor
- Weekly founder-level mentorship
- A real reference for graduate school applications or first AI safety roles
Summer or Semester Internship
You work alongside the founder on a live client engagement or a public artifact. Daily standup, real ownership of a piece of the work, and your name on what ships. A full year of AI assurance experience compressed into a single semester.
- A specific deliverable inside an active engagement such as eval suite design, scoring runs, or report drafting
- A public artifact like the next sectoral playbook or vendor scorecard
- Hands-on work with frontier models, evaluation harnesses, and regulatory frameworks
- A named contribution on a real signed deliverable
- Where the work yields a paper or a patent, the firm sponsors the submission or filing cost, and you are named as a co-author or inventor
- Direct access to how an independent AI assurance firm operates day to day
- A reference letter on Lattice/AI letterhead
The qualities that matter.
- A strong technical foundation in computer science, machine learning, security, mathematics, or statistics. A policy or economics background works too if you are seriously interested in AI governance.
- Curiosity about how AI systems actually fail in production, not just how they perform on a leaderboard.
- Willingness to publish under your own name and stand behind what you wrote.
- The ability to say “I don’t know” out loud, repeatedly, without it bothering you.
We are a small founder-led firm that just launched. We do not pay cash for these engagements yet, and we will not pretend otherwise. What we invest in is your published output. We cover the cost of getting your work into the public record: conference submission fees, journal article processing charges, and patent filing costs where the work is patentable. Your name appears on the paper, the patent, or the public scorecard as a co-author or inventor in your own right.
If you need paid work this summer, this is not the right fit. If you want a publication or a patent with your name on it, a real reference in the AI assurance space, and founder-level mentorship for the duration of the work, that is the deal we can offer today. We will revisit cash compensation as soon as the firm can support it, and alumni of these pathways will be first in line.
One short email is enough.
Email careers@latticeevals.com and tell us three things: who you are, one example of work you have shipped (a link is fine), and which track interests you. No formal CV is required. We respond within seven days.