Pranjal Aggarwal

Aggarwal completed undergraduate work at IIT Delhi, then spent research time at Princeton with Karthik Narasimhan before beginning a PhD in language technologies at Carnegie Mellon with Sean Welleck. His personal site frames the through-line as language models that act in the world: optimizing how they generate, how they use tools, and how content creators show up when those models answer questions.

With the GEO project, Aggarwal and coauthors asked whether content creators could measure and improve their visibility inside generative engines the way SEOs once measured rankings.

In November 2023 the team posted GEO: Generative Engine Optimization on arXiv. KDD 2024 later accepted the work. The paper treated visibility inside generative answers as a black-box optimization problem, defined metrics for how often and how prominently a source appears in generated responses, and showed that targeted content edits could raise visibility in their experimental setting.

Aggarwal is listed as lead author. Equal-contribution notes place Princeton PhD collaborators such as Vishvak Murahari and Ameet Deshpande beside him on measurement and benchmark design, with Narasimhan as senior academic and independent coauthors including Tanmay Rajpurohit and Ashwin Kalyan on the line.

The paper introduced GEO-bench, a public query set used to measure how content changes show up inside generative answers. In the study's setting, edits such as adding statistics, quotations, and citations could raise a source's visibility by large relative margins, with reported lifts up to about 40 percent depending on the method and domain.

Later commercial tools and practitioner studies often position themselves relative to GEO-bench even when they use different engines or prompts.