AirOverflow is venturing into one of the most exciting and consequential frontiers in modern cybersecurity — the security of Artificial Intelligence and Machine Learning systems — and we're looking for a sharp, intellectually driven intern to be part of this journey from the ground up. As AI and ML models become deeply embedded in critical applications, infrastructure, and decision-making systems worldwide, securing them has become not just important but absolutely essential. This is a field that sits at a rare and powerful intersection of two of the most complex disciplines in technology, and very few professionals globally — let alone in Pakistan — have developed meaningful expertise in it. As an AI/ML Security Engineer Intern at AirOverflow, you will explore how AI systems can be attacked, manipulated, and defended — from adversarial attacks and model inversion to prompt injection and data poisoning. You will work alongside our security team to research offensive and defensive techniques specific to AI/ML environments, contribute to internal tooling and research, and help AirOverflow build out a capability that is genuinely ahead of its time. We're not looking for someone who has it all figured out — we're looking for someone who is deeply curious, analytically strong, and excited about operating at the bleeding edge of both AI and cybersecurity. If that sounds like you, this internship could be the beginning of something remarkable.

A Note on Requirements: Our requirements for internship positions are not strict. The job description simply outlines what we would love for you to have — not a checklist you must fully satisfy. If you are curious, driven, and eager to learn, we want to hear from you.

What You'll Do

• Research and study AI/ML-specific attack techniques including adversarial examples, model inversion, membership inference, and data poisoning
• Assist in assessing the security posture of AI/ML pipelines and model deployment environments
• Explore prompt injection, jailbreaking, and other LLM-specific attack vectors and document findings
• Support the development of internal tools and frameworks for AI security testing and red teaming
• Review and analyze AI/ML models and their underlying infrastructure for potential vulnerabilities
• Stay current on academic research, CVEs, and emerging threats in the AI security space
• Produce clear technical writeups, research summaries, and documentation of experiments and findings
• Participate in team discussions and contribute ideas to AirOverflow's growing AI security practice

What We're Looking For

• Foundational understanding of Machine Learning concepts — supervised/unsupervised learning, neural networks, model training and inference
• Familiarity with Python and common ML libraries such as TensorFlow, PyTorch, or scikit-learn
• Basic awareness of cybersecurity principles and common attack/defense concepts
• Strong analytical and research skills with the ability to break down complex, unfamiliar problems
• Intellectual curiosity and self-motivation to explore a field that is still being defined
• Ability to read, understand, and summarize academic papers and technical research
• Clear and structured written communication for documenting findings and research

Nice to Have

• Exposure to adversarial machine learning concepts or tools such as CleverHans, ART (Adversarial Robustness Toolbox), or Foolbox
• Familiarity with LLMs and experience experimenting with prompt engineering or jailbreaking techniques
• Understanding of MLOps pipelines and how models are deployed in production environments
• Prior experience with penetration testing, CTFs, or any offensive security background
• Coursework or personal projects at the intersection of AI and security
• Familiarity with cloud platforms (AWS, GCP, Azure) and how ML workloads are hosted

What We Provide

• Mentorship at the crossroads of two of the most cutting-edge fields in technology
• Hands-on research experience in AI/ML security — a discipline with very few practitioners globally
• Access to AirOverflow's internal knowledge base, resources, and security research environment
• Internship certificate and letter of recommendation upon successful completion
• Opportunity to co-author research, contribute to publications, or present findings internally
• A potential pathway to a full-time role as AirOverflow scales its AI security capabilities
• The opportunity to be among the first in Pakistan to build a career at the frontier of AI and cybersecurity