Yash Vadi

I am currently pursuing a Master's degree in Computer Science with a specialization in Machine Learning at the Université de Montréal, where I am also affiliated with MILA.

I did my undergraduate studies at Indian Institute of Information Technology (IIIT), Surat in Electronics and Communication Engineering. During my undergrad, I interned at SHL Labs where I worked majorly on the problems in talent assessment for proctoring(detecting off-topic and AI/LLM-generated responses). Prior to SHL Labs, I worked under Dr. Anil Kumar (Scientist/Engineer - SG at IIRS, Indian Space Research Organization ), focusing on object segmentation using computer vision and fuzzy algorithms.

Feel free to drop me an email; I'm enthusiastic about exploring new areas(deep learning and NLP) and am always open to research collaborations.

Email  |  Resume  |  LinkedIn  |  Github

UPDATES

July 2023

We received an acceptance for our paper on Unsupervised Out-of-Distribution Dialect Detection with Mahalanobis Distance at INTERSPEECH 2023!

April 2023

Check out our blog on ChatGPT Detection: The Battle of the Pens: Insights into Detecting AI-Generated Text


RESEARCH EXPERIENCE

Researcher Intern
SHL Labs
Gurgaon, India (Jan 2023 - July 2023)

• Worked on LLM-generated text response detection for the Writing assessment proctoring.

• Integrated the Personalized Follow-up question generation module into the AI-based virtual interviewing platform.


Researcher Intern
SHL Labs
Gurgaon, India (Dec 2021 - Mar 2022 & May 2022 - July 2022)

• Worked on Out-of-Domain intent classification problem for the conversational ChatBot application.

• Developed the off-topic response detection module for the Writing assessment and created adversarial

testing pipeline to test the robustness of the essay grading model.

Research Intern
IIRS, Indian Space Research Organization (ISRO)

Dehradun, India (Sep 2020 - Nov 2020)
Advisor: Dr. Anil Kumar

• Trained an object detection model(SSD) for detecting infrastructures in the remote sensing satellite images.

• Utilized a fuzzy based clustering algorithm for the pixel-wise soft classification of the detected infrastructures.




RESEARCH PAPERS

Unsupervised Out-of-Distribution Dialect Detection with Mahalanobis Distance
Sourya Dipta Das, Yash Vadi, Abhishek Unnam, Kuldeep Yadav
[Accepted] INTERSPEECH 2023
Pdf


(website template modified from repo )