Rohit Prasad

In Amazon Unbound (Brad Stone), Prasad leads the deep-learning bet that becomes Alexa’s response model. The highlights credit him with explaining why Evi’s rigid knowledge graphs couldn’t be Alexa’s foundation — a graph-based system might mishear “Play music by Sting” as a goodbye to the artist — and why statistical deep-learning methods instead let the system infer that the intent is almost certainly to “blast ‘Every Breath You Take.’” Once Jeff Bezos “poured gasoline” on the early sparks, Prasad described the run-up to Alexa and Echo as a “running success” that changed his life; he was promoted to VP and eventually joined Amazon’s senior leadership (S-team), though he noted he had the launch playbook but “the playbook for the next five years, I didn’t have.”

Context: Rohit Prasad is an Amazon engineering leader who served as head scientist for Alexa and later as SVP and head scientist for Artificial General Intelligence (AGI) at Amazon.

Where this appears

• Amazon Unbound — drives Alexa’s shift from knowledge graphs to deep-learning response modeling; promoted to VP/S-team as Echo succeeds under Jeff Bezos’s experiment-then-scale playbook.