Lev Fedorov is a software engineer at Amazon specialising in large-scale recommender systems and ranking for consumer products. He has worked on short-video feeds, image search and discovery systems serving millions of users, owning everything from candidate generation and ranking models to experimentation frameworks and observability. Lev has led migrations from classic k-NN + matrix factorisation stacks to more interpretable architectures with per-request traces and author-facing dashboards, helping product and trust & safety teams understand “why this item was shown”. He regularly shares his experience at industry events and community meetups, with talks on building recommenders from scratch, tackling popularity bias and feedback loops, and making ranking more transparent and governed.