ML/AI Intern

About ATRISI

ATRISI builds institutional intelligence infrastructure for the AI-native era through applied research, enablement programs, and platforms including JoaLLM and TWAI. They ship real systems, experiments, and deployable workflows.

About the Role

This internship is part of the technical build track, focusing on hands-on ML and applied AI work supporting platforms, programs, and internal research, with mentorship and clear deliverables.

Role Summary

An ML / AI Intern will work on models, data, evaluation, and product-facing AI features within the ATRISI ecosystem. Responsibilities include running scoped experiments, documenting results, and prototyping capabilities such as RAG, retrieval, and workflow intelligence, with code review and guidance from senior builders. This is a builder internship on real codebases.

Key Responsibilities

• ML & Model Work
• Support training, fine-tuning, or evaluation on well-scoped tasks.
• Benchmark models and summarize tradeoffs (quality, latency, cost).
• Assist with dataset preparation, labeling guidelines, and data quality checks.
• Contribute reproducible notebooks or scripts (versioned, reviewable).
• AI Product Features
• Prototype RAG, retrieval, or agent-style workflows aligned with platform and program needs.
• Run structured evaluations: accuracy, failure modes, edge cases, regression checks.
• Assist in integrating model outputs into APIs or internal tools under senior review.
• Participate in code review and testing for AI-related changes.
• Research & Documentation
• Write short experiment summaries for engineering and program teams.
• Capture learnings in internal notes or knowledge products.
• Support applied research pilots requiring ML instrumentation.

How We Work

• Async updates via GitHub, Notion, or agreed tools.
• Join working sessions when useful.
• Raise blockers, data issues, and scope risks early.

Requirements & Tech Stack

• Must Have

• Pursuing or recently completed a degree in CS, AI, Data Science, or a closely related field.

• Strong Python and comfort with at least one of: PyTorch, scikit-learn, Hugging Face.

• Solid fundamentals: train/validation split, common metrics, overfitting, training vs. inference.

• Ability to read documentation or papers and ship a small working experiment.

• Git discipline: clear commits, reproducible runs, honest reporting of results.

• Genuine interest in LLMs, RAG, or applied AI in products.

• Strong Signals (Prioritized)

• A GitHub or portfolio with ML projects (coursework, capstone, Kaggle, side projects).

• A short write-up of one project: problem → approach → stack → learnings/deliverables.

• Evidence of caring about evaluation, not only demos.

• Comfort picking tools (e.g., Python, Jupyter, HF, SQL, Docker) and learning others on the job.

• Nice to Have

• Vector DBs, embeddings, or orchestration (e.g., LangChain-style) experience.

• FastAPI, Node, or light full-stack glue for wiring models to products.

• Builder-style work: deployed demo, RAG app, agent workflow, or Amplify-like capstone.

• Notebook, blog post, or open-source contribution.

Not the Right Fit If

• You only want generic “AI awareness” or prompt-playground work with no code.
• You expect exclusively live teaching or cohort operations.
• You cannot commit to regular async progress and documented deliverables.

What You Will Gain

• Mentorship on production-minded ML and AI feature work.
• Exposure to a multi-platform AI-native ecosystem.
• Portfolio artifacts tied to real initiatives.
• A credible path toward ATRISI Fellowship product engineering or research associate tracks for strong performers.

Logistics

• Employment Type: Internship (weekly availability discussed at apply).
• Location: Bengaluru, India.