Cyber & AI Security · Internship Role
AI RED TEAM
Stress-test chatbots and AI tools for harmful, wrong or unsafe behaviour, systematically and ethically, with reports developers can act on.
SKILLS YOU'LL WALK AWAY WITH
Adversarial testingJailbreak taxonomyAutomated scanning (garak, PyRIT)Harm categoriesResponsible disclosure
PROOF-OF-WORK TASKS8 TASKS
Start at Beginner and work up. Every task produces something real you can show: a report, a tool, a guide or a demo. Publish it on LinkedIn or GitHub as your proof of work.
B
Beginner
LEVEL 01
- 01Pick an AI chatbot you are allowed to test (your own, or an official challenge). Write a test plan with 5 harm categories, 5 prompts each, and what a 'fail' looks like.
- 02Test an AI tool's factual reliability: ask 20 questions with known answers in your subject and record every confident wrong answer.
- 03Write a one-page guide to responsible AI red teaming: scope, permission, what not to do, and how to report.
I
Intermediate
LEVEL 02
- 01Run garak or promptfoo red-team scans against a local open-source model. Interpret the results and manually verify the top findings.
- 02Test a chatbot in Hindi, Tamil or another Indian language and in English. Document whether its safety behaviour changes by language.
- 03Join a public AI red-teaming challenge (HackAPrompt or similar) and publish your approach and lessons after it ends.
A
Advanced
LEVEL 03
- 01Run a full red-team engagement on an AI app you built or have permission to test, using multi-turn attacks, role-play, encoding tricks and tool misuse. Deliver a findings report with severity ratings.
- 02Build an automated pipeline with PyRIT that generates, sends and scores attack prompts, and track the attack success rate across model or prompt versions.
HAND-PICKED FREE RESOURCES6 RESOURCES
FREE TOOL
promptfoo Red Teaming
Generates and runs adversarial tests against your AI app.
VISIT RESOURCE
FREE TOOL
HackAPrompt
Public AI red-teaming competitions and challenges.
VISIT RESOURCE
CODE REPO
garak
An open-source LLM vulnerability scanner.
VISIT RESOURCE
CODE REPO
PyRIT
Microsoft's open-source framework for automating AI red teaming.
VISIT RESOURCE
ARTICLE GUIDE
Microsoft AI Red Team Guidance
How one of the largest AI red teams plans and runs engagements.
VISIT RESOURCE
ARTICLE GUIDE
MITRE ATLAS
A knowledge base of real attacker tactics against AI systems.
VISIT RESOURCE