The Elephant in AppSec

My guest today is Petra Vukmirovic, Head of Information Security and IT at Numan, and she also works with DevArmor on automating threat modeling and security design reviews.
Outside of that she started the OWASP Threat Model Library, an open collection of real threat models the community can learn from.

What makes her path unusual is that she didn't come to AppSec through development, she came through emergency medicine, where she worked as a doctor.

In this episode, we talked about what transfers from the ER to incident response, which is mostly the protocols: risk scores, runbooks, decision trees you can follow when things are on fire. She also thinks threat modeling stops too early. Most teams model protective controls and stop, when recovery deserves the same attention. 
We also got into automating threat models with LLMs, catching drift between the model and the code, and where design reviews end and threat modeling begins.
And much more!
This podcast is brought to you by
Escape: https://escape.tech  — Offensive security for the teams that are 100x outnumbered, combining Attack Surface Management, business-logic-aware DAST and AI pentesting solutions.

Today I'm joined by Advait Patel, Senior Site Reliability Engineer and the creator of DockSec, an open-source, AI-powered Docker security scanner that's now an official OWASP Incubator project.

In this episode, we get into:
Why dumping 200 container findings into a Jira ticket is the fastest way to get developers to fix nothing and how DockSec cuts that down to the 5 that actually matter

The AI support agent that got hijacked by a single malicious ticket and emailed customer data straight to an attacker

Why you should treat AI as an assistant on a leash, not an engineer with root access

the Docker mistakes Advait sees everywhere (stale base images, root by default, and secrets baked right into the image)

…and much more!
Get ready, Advait doesn't hold back his opinions. Let's dive right in!
Connect with Advait: https://www.linkedin.com/in/advaitpatel93/
Connect with Alexandra: https://www.linkedin.com/in/alexandra-charikova/
This podcast is brought to you by
Escape: https://escape.tech  — Offensive security for the teams that are 100x outnumbered, combining ASM business-logic-aware DAST, and AI-powered pentesting solutions.
Mentioned
DockSec on GitHub (now the OWASP org repo): https://github.com/OWASP/DockSec
OWASP project page: https://owasp.org/www-project-docksec/
Open Policy Agent (his "open policy" reference): https://www.openpolicyagent.org/
OWASP Top 10 for LLM Applications: https://genai.owasp.org/

Today, I'm joined by Kavia Venkatesh, Director of Product Security at a large healthcare organization. She didn't take the traditional path into cybersecurity — she came from biotech. But that outsider lens turned out to be her edge.

With over 10 years of experience leading cybersecurity strategy for hyper-scale ecosystems, she's built many security programs from the ground up, navigating 9 acquisitions in 18 months at a large tech org, and along the way developed a rare ability to translate risk into language that executives actually act on.
Kavia is also a frequent speaker at premier global conferences, including DEF CON, BSides San Francisco, and Nullcon.

In this episode, we talked about what most security teams get completely wrong during integrations, what she'd change about how security teams show up in organizations and the "breachability mindset" that changes how you approach risk.
And much more!

Get ready, Kavia doesn't hold back her opinions. Let's dive right in!

This podcast is brought to you by Escape: https://escape.tech — Offensive security for the teams that are 100x outnumbered, combining Attack Surface Management, business-logic-aware DAST and AI pentesting solutions.

Connect with Kavia: https://www.linkedin.com/in/kaviavenkatesh/

Today, I’m joined by Jason Fernandes, VP of security and privacy at Mercari, the Japanese-born global marketplace now spanning e-commerce, FinTech, and crypto. It is this rare combination that puts him at the intersection of some of the strictest regulatory environments in tech.
He oversees everything from product and platform security to threat detection, privacy, and, since last year,  AI security and AI governance.
In this episode, we also talked about the challenges of AI governance, the lethal trifecta for AI agents, the confused deputy problem, and how to justify AI security investments to the leadership and working with FinOps teams. And much more!
Dive right in!
This podcast is brought to you by
Escape: https://escape.tech  — Offensive security for the teams that are 100x outnumbered, combining Attack Surface Management, business-logic-aware DAST and AI pentesting solutions.
Mentioned
FACADE (Google's internal fraud detection model) https://arxiv.org/abs/2412.06700
Meta Practical AI Agent Security (Rule of Two) https://ai.meta.com/blog/practical-ai-agent-security/
Simon Willison The Lethal Trifecta https://simonwillison.net/2025/Jun/16/the-lethal-trifecta/
Hiroki's AI Security blog (Mercari) https://hi120ki.github.io/blog/posts/20260103/
Anthropic  Project Vend https://www.anthropic.com/research/project-vend-2

Today, I’m joined by Amol Deshpande, a seasoned security engineer currently at Stripe, where he focuses on building secure systems at massive scale. With a background spanning product security and penetration testing at companies like Salesforce, Splunk, and Early Warning, Amol brings deep hands-on experience in securing complex, real-world platforms.
He’s also been a HackMIT judge and a long-time CTF competitor at DEF CON, giving him a very practical view of modern security challenges.
In this episode, we cover whether security must now belong in every AI strategy meeting, and how to embed it into AI development from the outset.
We also touch on how privacy concerns will only grow as agents are trained on sensitive data and why human oversight is essential for critical AI operations.
Dive right in!