Agenda:6:00 - 6:15pm: Introduction
6:15 - 7:00pm: Girish Chandrasekar - Benchmarking the Offensive Security Capability of Foundation Models
7:15 - 8:00pm: Networking
Summary of the talk:
In this talk, Girish will present the methodology and results of a benchmark designed to assess whether frontier AI models and agentic coding tools materially improve offensive cyber capabilities. The benchmark uses a homegrown evaluation tool built on three production-grade cyber ranges—Bank, Hospital, and Nuclear Reactor—to compare skilled human operators working with and without AI assistance. It measures uplift across success rate, speed to objective, minimum skill threshold, stealth, and estimated real-world harm in terms of dollars and lives.
Speaker bio:
Girish Chandrasekar is Head of Product at Straiker, an AI-native security platform protecting agentic AI applications. He leads product strategy across Straiker's Defend, Ascend, and Discover offerings. Previously, Girish was AI Product Lead at Robust Intelligence (now acquired by Cisco), and held roles on the AI/ML teams at Postmates and JP Morgan. He holds a BA from Columbia University. He is based in the San Francisco Bay Area.
Agenda:6:00 - 6:15pm: Introduction
6:15 - 7:00pm: Girish Chandrasekar - Benchmarking the Offensive Security Capability of Foundation Models
7:15 - 8:00pm: Networking
Summary of the talk:
In this talk, Girish will present the methodology and results of a benchmark designed to assess whether frontier AI models and agentic coding tools materially improve offensive cyber capabilities. The benchmark uses a homegrown evaluation tool built on three production-grade cyber ranges—Bank, Hospital, and Nuclear Reactor—to compare skilled human operators working with and without AI assistance. It measures uplift across success rate, speed to objective, minimum skill threshold, stealth, and estimated real-world harm in terms of dollars and lives.
Speaker bio:
Girish Chandrasekar is Head of Product at Straiker, an AI-native security platform protecting agentic AI applications. He leads product strategy across Straiker's Defend, Ascend, and Discover offerings. Previously, Girish was AI Product Lead at Robust Intelligence (now acquired by Cisco), and held roles on the AI/ML teams at Postmates and JP Morgan. He holds a BA from Columbia University. He is based in the San Francisco Bay Area.
Agenda:6:00 - 6:15pm: Introduction
6:15 - 7:00pm: Girish Chandrasekar - Benchmarking the Offensive Security Capability of Foundation Models
7:15 - 8:00pm: Networking
Summary of the talk:
In this talk, Girish will present the methodology and results of a benchmark designed to assess whether frontier AI models and agentic coding tools materially improve offensive cyber capabilities. The benchmark uses a homegrown evaluation tool built on three production-grade cyber ranges—Bank, Hospital, and Nuclear Reactor—to compare skilled human operators working with and without AI assistance. It measures uplift across success rate, speed to objective, minimum skill threshold, stealth, and estimated real-world harm in terms of dollars and lives.
Speaker bio:
Girish Chandrasekar is Head of Product at Straiker, an AI-native security platform protecting agentic AI applications. He leads product strategy across Straiker's Defend, Ascend, and Discover offerings. Previously, Girish was AI Product Lead at Robust Intelligence (now acquired by Cisco), and held roles on the AI/ML teams at Postmates and JP Morgan. He holds a BA from Columbia University. He is based in the San Francisco Bay Area.
Speakers
Girish Chandrasekar
