Kshitij Gupta, MSc in Applied Computing from the University of Toronto, is a Machine Learning Engineer at Meta. His work focuses on production machine learning and AI systems, including recommendation systems, semantic retrieval, embedding-based search, ranking infrastructure, feature pipelines, and evaluation frameworks.

Across consumer technology, life sciences, emissions modelling, and security infrastructure, Kshitij has focused on turning research ideas and fast-moving prototypes into reliable production systems. His perspective is shaped by environments where quality is not just a model score or a demo result, but an engineering responsibility carried through architecture, deployment, monitoring, and ownership.

At Ai4 2026, Kshitij will present "Cheap to Write, Expensive to Trust: Architecting Production Systems with LLMs." The talk examines what it takes to use LLMs in real software without handing them unchecked authority. He will discuss how teams can preserve trust by separating deterministic cores from generative shells, adding verification gates and evals, and keeping guarantees owned by humans rather than inferred from model behaviour.