AI Research Intern
Work on assessment AI or neural-symbolic policy learning alongside the team, with a real chance of co-authorship if the work earns it.
- Team
- Research
- Type
- Internship
- Location
- Remote · India
- Duration
- 6 months
- Compensation
- ₹20,000 – ₹35,000 / month
What you will actually do
You will own a research question, not a ticket queue. In the first month you will read the surrounding literature and write a one-page problem statement that survives a review. After that, you run experiments, keep an honest record of what failed, and write up what holds.
Two of our current directions:
- Assessment AI — how a grading model explains its reasoning to a faculty member well enough that they can disagree with it precisely.
- Neural-symbolic policy learning — robotic control that fuses reinforcement learning with symbolic constraints, because a policy you cannot explain is a policy you cannot deploy.
What we are looking for
- Working Python, and comfort with PyTorch or JAX
- Linear algebra and probability you can use, not just recall
- Evidence you have implemented a paper’s method from its description
- The specific stubbornness required to debug an experiment for three days
You do not need publications. Several of our interns arrived with none and left as first authors.
How we work
Weekly one-to-one with your supervisor, a reading group on Fridays, and a written research log that you keep and we read. No standups. No performative busyness. We care what you found out this week.
Applying
Send us something you have built or written — a repo, a report, a blog post, a failed experiment you understood well. A CV alone tells us very little.
Show us something
you made.
A repo, a report, a deployed thing, a write-up of an experiment that did not work. A CV alone tells us very little, and we would rather judge you on the work.
Apply — AI Research Intern
Links first. We read them.