Jeet started in computer networking before making the switch to offensive security as an ethical hacker consultant running penetration tests across a wide range of industries and attack surfaces.
Now he works as an Offensive Security Engineer at Robinhood, where the attack surface is unlike anything in traditional enterprise security. Financial systems. Real money. Complex multi-layered codebases built to move fast.
His current focus: figuring out where AI and LLMs actually help in offensive security, and more importantly, where they fall short. Not the vendor pitch version. The version you get when you actually put these tools up against real production code.
Talks at DC416
LLMs Don't Hack, They Guess — Lessons from Building Offensive AI for Multi-Layered Codebases
AI tools can find the obvious bugs. They follow the most suspicious-looking signal until it checks out, then move on. The boring parameter — the one that actually leads somewhere — often gets skipped because it never looked dangerous enough to explore deeply. This talk breaks down the structural reasons AI agents fail at exhaustive vulnerability discovery, where they genuinely shine, and what a better approach looks like when you combine human intuition with machine speed.
Affiliations
- Robinhood — Offensive Security Engineer
- Ethical Hacker Consultant — Penetration Testing (prev.)
- Computer Networking — Industry background