Vasily Starostenko AI Safety
The problem: AI is moving into the real world faster than safety engineering can keep up. Companies ship autonomous and semi-autonomous systems, then discover the hardest failures are human: operators who lose vigilance, supervisors who lose situational awareness, alerting systems that cry wolf. Safety gets bolted on after launch — usually after an incident.
Why Vasily: he has patented answers to these exact problems. As a product leader for driverless fleet operations at Waymo, he is a co-inventor on four granted US patents covering the safety layer around human monitors of autonomous vehicles: psychomotor vigilance testing (US 11,826,146), situational awareness testing (US 12,024,206), head-movement monitoring (US 11,912,307), and fatigue monitoring (US 11,738,763). This is not commentary on AI safety — it is issued intellectual property, invented while operating a driverless fleet across multiple cities.
What you get: an AI safety assessment that maps your system's real failure modes — monitoring gaps, escalation paths, and the moments nobody owns; a human-factors review of how your operators or users actually stay alert inside an automated system; and ongoing advisory as you design safety-critical features. Safety as a system, not a feature.
The proof: two years leading Waymo's driverless fleet operations through research-to-commercialization, four granted patents in autonomy safety, and twenty years of product leadership where reliability was never optional. A regular speaker on automation, he thinks about the watching with the same rigor most teams reserve for the machine.
The stakes are rising beyond vehicles. The same human-factors problems Vasily patented solutions for — vigilance decay, lost situational awareness, fatigue in the monitor — now apply to every AI system with a human in the loop: copilots, agents, automated operations centers. A regular speaker on automation, he brings a practitioner's skepticism to safety theater. Real safety is measured in failure modes understood, escalation paths tested, and systems that get safer with every incident. If your AI touches the real world, that rigor is not optional — and it cannot be retrofitted.
Contact Vasily Starostenko Phone: +1 (650) 523-4953 — tap the call button on this page to reach him directly. Priority scheduling ($2,000 to speed up scheduling): https://buy.stripe.com/aEUdTVbwtfjn5A49AA Every engagement is confidential, tailored, and focused on measurable outcomes.
Vasily Starostenko — Consulting
Talk now — free call