Research and Software Development Intern (AI)

Responsibilities:
Reading and stress-testing recent papers against real-world problem constraints.
Designing and running experiment ablations, benchmarks, failure-mode analysis, not just implementing a paper's happy path.
Prototyping candidate architectures and defending design choices with evidence, not intuition.
Turning validated experiments into documented, reproducible solution components.
Writing technical rationale that holds up under review: what was tried, what failed, why the final approach won.
Ramp Plan:
30 days: Fluent in the problem space and existing experiment infrastructure. Running your first independent experiment.
60 days: Owning a research question end-to-end literature review through experiment through documented result.
90 days: Your experimental findings are shaping an actual architecture decision, not sitting in a notebook.
Requirements:
Pre-final or final year, CS/AI/ML or related field.
Strong Python fundamentals.
Can read a paper critically and identify what's actually novel vs. incremental.
Comfortable designing experiments, not just running them; knows what a fair ablation looks like.
Comfortable with ambiguity and being wrong in public (research means most experiments don't work).
Nice-to-Have:
PyTorch/transformers experience.
Prior exposure to VLM, or self-supervised architectures.
Open-source contributions or competition history.
Interest in geospatial, sensor, or industrial time-series data.

Experience
0-0 yrs

Employment type
Internship