I'm Yasir khan
About Me
A brief look into who I am, what drives my research, and what I'm looking for next
Background
I'm a Machine Learning Engineer and Security Researcher with a passion for building intelligent systems that are both powerful and resilient against adversarial threats. My journey has taken me through competitive cybersecurity — placing in multiple national CTF competitions — while simultaneously deepening my expertise in applied machine learning. I've completed my bachelor's studies in Computer Science from Abdul Wali Khan University Mardan with an impressive ranking of top 3% in the class.
AI Engineering
I design and build intelligent AI applications that solve real-world problems, with a focus on
Generative AI, LLMs, RAG, AI agents, and workflow automation.
Generative-AI & LLMs
Building production-grade applications powered by large language models — from fine-tuning
and prompt engineering to designing reliable, scalable inference pipelines that turn LLM
capabilities into real product features.
RAG & Multi-Agent Systems
Building retrieval-augmented and multi-agent architectures that combine grounded knowledge with
coordinated, autonomous reasoning. I design RAG pipelines that retrieve accurate, up-to-date context
to reduce hallucination, and multi-agent systems where specialized agents plan, delegate, and verify
each other's work.
AI Agents & Workflow Automation
Developing autonomous and semi-autonomous agents that plan, use tools, and complete
multi-step tasks — automating complex workflows and integrating AI into real business
processes end to end.
Research Interests
My research spans machine learning security, adversarial robustness, and applied cryptography —
with hands-on work across competitive cybersecurity, threat detection systems, and AI safety.
Adversarial Machine Learning
Studying how machine learning models fail under adversarial pressure, and designing defenses
that hold up against evasion, poisoning, and model-extraction attacks. I'm interested in the
trade-offs between model robustness, accuracy, and real-world deployability.
Privacy & Security in LLMs
Investigating how large language models can be compromised — through prompt injection, jailbreaking,
or leakage of sensitive training data — and researching defenses that keep them safe to deploy. My
focus spans robust alignment techniques, input/output sanitization, and privacy-preserving methods for
training and inference that protect both users and the underlying data.
AI for Cybersecurity
Applying machine learning to strengthen defensive security operations — from detecting anomalies and
intrusions in real time to automating vulnerability discovery and threat triage. I'm interested in how
AI systems can proactively identify emerging attack patterns and reduce response time, while remaining
robust enough to avoid becoming an attack surface themselves.