top of page

Workshop

Sagar Bhure

Sr Software Engineer

F5

Sagar Bhure (https://sagarbhure.com/) is a highly accomplished Security Researcher with a proven track record of excellence in his research on security. He is a filed patent holder with the US for his innovative work on ML and Security and has published several papers on the subject in top-tier journals. He currently leads various projects at OWASP, including the prestigious 'ML Security Top 10' ,an OWASP flagship project. Sagar has spoken at several industry-leading international conferences, including Hack in Paris, BlackHat, BSides, OWASP, and APISecure. He is regarded as a respected thought leader in the cybersecurity community, frequently invited to speak at conferences and workshops on topics related to offensive and defensive security. Sagar’s engaging presentations have helped to educate security professionals with cutting-edge research and tools to strengthen their security toolkits.

Machine Learning For Security Professionals: Building And Hacking ML Systems

Our training offers an easy-to-follow introduction to machine learning for security professionals, even if they have no prior math or machine learning knowledge. In the ML4SEC part, attendees will get hands-on experience creating both defensive and offensive security tools using popular libraries like TensorFlow, Keras, PyTorch, and scikit-learn. We'll guide you through the entire machine learning process, from preparing data to building, training, evaluating, and using ML models. In the SEC4ML section, we'll delve into vulnerabilities in advanced machine learning methods, including adversarial  learning, model stealing, data poisoning, and model inference. Participants will work with vulnerable machine learning applications to understand these weaknesses thoroughly and learn how to mitigate them. Our training equips security professionals with practical knowledge they can directly apply in their work.

Additionally, participants will explore recent advancements in creating highly realistic images of non-existent people and the associated risks for fraud detection and disinformation campaigns. Through hands-on exercises and real-world examples, they will learn how to identify vulnerabilities in image forensic classifiers and understand how these attacks can severely impact accuracy.

bottom of page