Engineers Clinic
AdvancedSoftware trackUS-CP-02

LLM Evaluation & Safety Harness

Golden datasets, LLM-as-judge scoring, adversarial red-teaming and CI merge gates for an LLM feature.

10 Guided Steps480h Internship TierSubmit: GitHub repo or evidence pack + live demo / report + README + milestone logbookVerified CertificateAI Rubric Graded
Program Overview

Golden datasets, LLM-as-judge scoring, adversarial red-teaming and CI merge gates for an LLM feature.

What you submit

GitHub repo or evidence pack + live demo / report + README + milestone logbook

How it is graded
  • Works as specified30

    Every feature in the brief runs end to end with real data, and the reviewer can follow the README to try it.

  • Code quality & design20

    Clear structure, sensible names, input validation, no secrets in the repo, and a short note on the main design choices.

  • Testing & evidence20

    Automated tests for the core paths, plus a live URL, screenshots or a short demo video.

  • Documentation15

    README covers purpose, setup, how to run tests, and known limits.

  • Reasoning & presentation15

    The write-up explains trade-offs and results in plain language, with numbers where they exist.

Pass 60% · Merit 75% · Distinction 90%

Learning Outcomes
480-hour United States programme. Evaluation is the scarcest skill in the AI stack and the hardest to fake in an interview.
Target Audience

Engineering Students

Diploma, B.Tech, B.E., BCA, MCA, or M.Tech students seeking practical experience.

Aspiring Developers

Developers who want to build proof-of-work capstone projects for tech resumes.

Job Seekers

Graduates looking for AI-evaluated code review & verified industry credentials.