Pranjal Rawat

Applied Scientist, Uber · PhD, Georgetown Economics

Hi, I'm Pranjal

I'm an incoming Applied Scientist at Uber working on algorithmic pricing in the Delivery Marketplace. I have a PhD in Econometrics and Quantitative Economics from Georgetown University and three years of full-time work experience in applied and data science (Amex, RecSys).

My research interests lie in ML-powered scalable methodologies for causal inference & experimentation, choice & preference modelling, reinforcement learning & dynamic programming, and optimization problems.

During my PhD, my advisors were John Rust and Nathan Miller and I also interned/worked with the following tech firms: Roblox, Uber and Topsort. Before my PhD, I worked as an Assistant Manager (Data Science) at American Express for 3 years on productionalizing marketing rec-sys and graph embedding systems. Before that, I worked on time series econometrics during my first Masters and at the Reserve Bank of India.

A Deep Learning Approach to Heterogeneous Consumer Aesthetics in Fast Fashion

FashionCLIP embeddings and a demand model on millions of H&M purchases. Recovers price and taste sensitivity.

arXiv PDF Code

Who is more Bayesian: Humans or ChatGPT?

Humans vs. ChatGPT on binary classification. Documents bias and version drift. With John Rust, Chengjun Zhang, Tianshi Mu, Aaron Zhong.

arXiv PDF

Algorithmic Collusion in Auctions

Controlled lab experiments measuring bid suppression across Q-learning, bandits, and pacing. WEAI 2025.

arXiv PDF Code

Approximating Auction Equilibria with Reinforcement Learning

PPO learns bidding strategies in multi-item and dynamic auctions.

arXiv HTML

A Survey of Reinforcement Learning for Economics

RL methods for economic problems with high-dimensional states and continuous actions.

arXiv PDF Code

Structural Econometrics and Reinforcement Learning

Synergies across finance, IO, and policy. Oxford Research Encyclopedia, with John Rust.

Notes on economics, machine learning, and whatever I'm reading. Subscribe →

'26

Jun

Remembering Dimitri Bertsekas

'26

May

Sharing some experiences on the Hiring of PhD Candidates

'26

Apr

Software for Economic Modelling

'26

Mar

I got my PhD, now what?

'25

Nov

How food is delivered

'25

Nov

How do platforms measure the effectiveness of ads?

'25

May

How to work with LLMs

'25

Feb

The Flippin Coin: A Short Story

'25

Jan

Econ 101: Who will Generative AI displace?

'25

Jan

Why are markets are so efficient?

'24

Oct

Solving Auctions with Reinforcement Learning

'24

Sep

Improving the precision of average treatment effects

TechEcon live site →

A Netflix-like website and digital library of resources for Econ PhD students, early-career economists and data scientists working in US tech industry. This aggregator site contains: a) Econometrics, Statistics, & ML packages (Python, R), b) Large industrial datasets from real firms, c) Learning resources, "getting started" roadmaps, d) YT series & podcasts, bloggers, events & conferences near you.

DeepInference

A Python package that implements the deep-learning-based estimators from Farrell, Liang, and Misra (2021, 2025), which combines the interpretability of classic statistical models (Logit, Poisson, etc.) with the representational power of deep nets, allowing parametric inference and capturing rich HTEs over high-dim covariates, images, text, sequences, and more.

PrefGraph

A Python package that implements revealed preference graph algorithms from leading papers in Microeconomics and Optimization to compute consistency and exploitability scores, and recover reward/utility functions from user choices from budgets, menus, lotteries, comparisons, etc.

EconIRL

A Python package that implements SOTA dynamic discrete choice and inverse reinforcement learning algorithms to help us understand and predict users' sequential choices.

Interactive Pricing Theory live site →

18 interactive tracks teaching pricing and revenue optimization through visualizations, full mathematical formulations, and hands-on simulators, drawing on Phillips' Pricing and Revenue Optimization.

econometrics-in-python ★ 61

A guide to the econometrics package ecosystem in Python, mapping the tools for each method to working code.

industry-datasets ★ 54

A curated catalogue of large datasets released by real firms, for empirical research and teaching.

computational-economics ★ 40

Computational methods and notebooks for solving richer economic models than closed-form theory allows.