Secure Sips

Name: Secure Sips
Start: 2025-08-02T11:00:00+05:30
End: 2025-08-02T14:00:00+05:30
Location: CloudSEK

Aug 2, 5:30 – 8:30 AM (UTC)

Join us on August 2nd for an exciting LLM-themed Secure Sips x Hacktron meetup in Bengaluru. Discover how Mohan (S1r1u5) used Offensive AI to break Windsurf & Cluely while building Hacktron, learn from pwnfunction as he shares how to build powerful infrastructure for AI agents , Pranav would teach us Bitter lesson for faster kernels, auth and agents by Sid and Mr rc on Using LLMs as assemblers

Educational WorkshopsHacking MeetupsStudent Series

About this event!

Mark your calendars for August 2nd and join the Secure Sips x Hacktron Meetup, featuring deep-dive sessions on AI and Security.
Kick off with Mohan (S1r1u5) as he demonstrates how Offensive AI can break tools like Windsurf & Cluely while building Hacktron, one of the smartest AI-driven security platforms.
Follow up with pwnfunction’s session on crafting scalable infrastructure for AI agents, offering a behind-the-scenes look at powering next-gen AI tools.
Then, Paranav takes the stage for a Bitter lesson for faster kernels  focusing on matmuls – the mathematical backbone of AI performance optimization.
Sid will give a talk on Auth and agents
and finally Mr Rc will give a talk on Using LLMs as nondeterministic assemblers
This meetup promises cutting-edge insights into how builders are breaking, making, and accelerating AI in today’s fast-evolving landscape.

Speakers

Mohan S1r1u5

Hacktron

Suraj Pwnfunction

Concave-dev

Pranav Deepak

siddharth balyan

Mr Rahul

When

Saturday, August 2, 2025
5:30 AM – 8:30 AM (UTC)

Where

CloudSEK
16/1 Cambridge Road
Bengaluru, 560008

Agenda

5:30 AM	Offensive AI can break tools like Windsurf & Cluely while building Hacktron	Offensive AI is redefining the security landscape, demonstrating how AI-driven attacks can break tools like Windsurf and Cluely with precision. At the same time, the same AI capabilities are being used to build Hacktron one of the smartest AI-powered security platforms ever made. This talk explores the dual nature of AI, where it acts as both the attacker and the ultimate defender. Step into the future of cybersecurity, where the battlefield is entirely AI-driven
6:15 AM	Building Infra for AI Agents	Behind every powerful AI agent lies a rock-solid infrastructure designed to handle scale, speed, and reliability. This session breaks down the foundations of building high-performance systems for modern AI workloads. From compute orchestration to data pipelines and secure execution layers, we’ll cover what truly powers next-gen agents. If you want to make your AI agents faster, smarter, and unstoppable, this is where you start
7:00 AM	Bitter lesson for faster kernels	Modern AI agents are only as fast as the math behind them. This session dives deep into GPU kernel optimization, focusing on matmul (matrix multiplications) the backbone of most AI computations. We’ll explore techniques to squeeze every bit of performance out of GPUs, from kernel-level tweaks to large-scale infra design. If you’re building AI systems and care about speed, this is where the magic happens.
7:30 AM	agents and auth -- exploring a smooth and secure auth experience for ai agents	AI agents are becoming more autonomous, but with power comes risk. This session dives into how agents authenticate, delegate access, and interact securely with APIs and systems. We’ll explore where today’s auth models break when applied to agents and what safer designs could look like. If you’re building or breaking AI agents, this is a must-watch
8:00 AM	Using LLMs as nondeterministic assemblers	Rahul (RC) has found out during his research that LLMs are quite good at generating direct binary code for simple x86 assembly snippets with some accuracy. The question is, can it be scaled?

Organizers

Akshansh Jaiswal

CRED

akshansh

See bio

charan Tadikonda

Hackerone

0xcharan

See bio

Contact Us