Watch the stream
When you're protecting a billion-user platform, attackers don't wait. Scraper bots, fake account farms, and credential stuffing campaigns operate at machine speed and your defenses need to be faster. This talk dissects the architecture of sub-millisecond anti-abuse detection systems that must make security decisions without adding perceptible latency to legitimate users.
We'll examine real-world defensive infrastructure using Venice (LinkedIn's derived data platform) as a case study, revealing the architectural patterns, trade-offs, and failure modes of ultra-low latency security systems. You'll learn how embedded data stores enable <1ms threat intelligence lookups, how precomputed reputation scores defend against distributed attacks, and critically—where the weaknesses lie.
Manu Jose Senior Manager, LinkedIn
With over 20+ years of software engineering experience, I am a technically oriented, high-energy, and empathetic leader who is passionate about building scalable, reliable, and innovative solutions for machine data. I am currently a Sr. Manager at LinkedIn, where I lead the Venice Project, a cutting-edge initiative that leverages online deep learning to improve user experience and engagement on the platform. I am driven by the mission of creating economic opportunity for every member of the global workforce, and I value collaboration, diversity, and continuous learning in my team.