Paridhi Maheshwari

Paridhi Maheshwari is an applied scientist at Alexa AI, Amazon, and a coauthor of the 2023 paper that introduced DSPy, a programming model for building and optimizing language-model pipelines. She finished a degree in electrical engineering at IIT Kanpur in 2018, then worked as a Research Associate at Adobe Research in Bangalore, where her papers on color and scene-graph retrieval appeared at SIGIR 2020 and AAAI 2021. At Stanford, where she started a master's in computer science in 2021, she was a research assistant in the SNAP Lab under Jure Leskovec and graduated with a distinction in research. In 2022, she and Sharmila Reddy Nangi built TeamOtter, a SemEval system for detecting misogynous memes that ranked 5th on Task A. Her DSPy coauthors included Omar Khattab, Matei Zaharia and Christopher Potts, and the paper was published at ICLR 2024; the 2024 TimeGraphs paper with Jure Leskovec and colleagues proposed a hierarchical temporal graph for reasoning about dynamic interactions.

Now at: Amazon
Works: 17
Citations: 122
h-index: 5
Badges: Bridge

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

2018

Electrical engineering at IIT Kanpur

In 2018, she completes a bachelor's degree in electrical engineering at the Indian Institute of Technology Kanpur, where Prof. Rajesh Hegde advises her.
2018–2021

Research Associate at Adobe Research

She joins Adobe Research in Bangalore as a Research Associate in 2018, having first come to Adobe as an intern. Until 2021, she works on learning associations between language and color and on using scene graphs for retrieval.
2020–2021

Color and scene-graph retrieval papers

Her paper "Learning Colour Representations of Search Queries" is accepted at SIGIR 2020. In 2021, she publishes "Scene Graph Embeddings Using Relative Similarity Supervision" at AAAI with Ritwick Chaudhry and Vishwa Vinay, using a graph convolutional network to turn scene graphs into image embeddings for semantic image retrieval.
2021

Stanford master's and the SNAP Lab

In 2021, she starts a master's in computer science at Stanford with a specialization in artificial intelligence. As a research assistant in the SNAP Lab under Prof. Jure Leskovec, she works on time-aware graph machine learning, and she graduates with a distinction in research.
2022

Detecting misogynous memes at SemEval

In 2022, she and Sharmila Reddy Nangi describe TeamOtter, their system for SemEval-2022 Task 5 on detecting misogynous content in memes. It pairs a multimodal BERT architecture with knowledge from pretrained CLIP and Urban Dictionary, and it ranks 5th on Task A with an F1 score of 0.679.
2023

Coauthoring the DSPy paper

In October 2023, she coauthors "DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelines" with Omar Khattab, Matei Zaharia, Christopher Potts and nine other authors. The paper introduces DSPy, a programming model that abstracts language-model pipelines as text transformation graphs, with a compiler that optimizes any pipeline to maximize a given metric, and it appears at the Twelfth International Conference on Learning Representations in 2024.
January 2024

TimeGraphs for temporal reasoning

In January 2024, she posts "TimeGraphs: Graph-based Temporal Reasoning" with Hongyu Ren, Yanan Wang, Rok Sosic and Jure Leskovec. The method models dynamic agent interactions as a hierarchical temporal graph and is evaluated on a football simulator, the Resistance game and the MOMA human activity dataset.
As of October 2026

Applied scientist at Alexa AI

In November 2021, she announces an internship at Amazon Alexa AI for the summer of 2022. She is an applied scientist at Alexa AI, Amazon.