Pranjal Chitale
Research Fellow
About
I am a Research Fellow at Microsoft Research India, where I work on advancing retrieval models and building multilingual AI systems.
Previously, I completed my MS (by Research) from the Indian Institute of Technology, Madras, where my thesis work led to the development of IndicTrans2 (opens in new tab) – the first open-source models supporting all 22 scheduled Indian languages with over 5M downloads on HuggingFace (opens in new tab) and deployed at the Honorable Supreme Court of India and WikiMedia foundation.
My current research interests encompass developing rigorous evaluation methodologies that reveal model limitations, data and resource-efficient learning, and understanding how to effectively and efficiently utilize synthetic data as a tool to make models more robust to out-of-distribution settings and its potential to unlock multilingual capabilities of models
Featured content
CVQA: Culturally-diverse Multilingual Visual Question Answering Benchmark
In this episode, Research Fellow Pranjal Chitale joins host Gretchen Huizinga to discuss the paper “CVQA: Culturally-diverse Multilingual Visual Question Answering Benchmark,” an oral presentation at this year’s Conference on Neural Information Processing Systems (NeurIPS). CVQA, which comprises questions and images representative of 31 languages and the cultures of 30 countries, was created in collaboration with native speakers and cultural experts to evaluate how well models perform across diverse linguistic and cultural contexts, an important step toward improving model inclusivity.