Sridhar Ragupathi

Seeking Machine Learning roles

Biography

I am a grad student in Professional Computer Science at Simon Fraser University and my interests are in Deep Learning for Computer Vision and NLP. Stumbling upon this video on neural networks as a freshman on summer break shaped my career.

Currently, I am using Computer Vision models to inspect defects and features on tires at 3DM Devices Inc. On my journey of exploring deep learning, I obtained versatile experience in Computer Vision and Natural Language Processing through both industry and research positions. In my previous stints, I was a Data Intern at Eltropy, forecasting text messages for Credit Unions in the US and I was an NLP Engineer at LimeChat, boosting sales and automating customer support for D2C brands.

I graduated from NIT Trichy with a Bachelors in Computer Science and Engineering. Through my undergraduate years, I was guided by bright minds such as Prof. Liming Chen, Prof. CV Jawahar, Prof. Vinay Namboodiri. I am grateful for receiving the Charpak Lab Scholarship which enabled me to pursue a research internship in France and the SPDC scholarship which funded part of my undergraduate education.

Download my resumé.

Interests

Deep Learning
Computer Vision
Natural Language Processing
Software Development

Education

MS in Professional Computer Science, 2024

Simon Fraser University
B.Tech in Computer Science & Engineering, 2020

National Institute of Technology Tiruchirappalli (NIT Trichy)

Skills

Python

C++

PyTorch

Shell Scripting

$brands/latex-logo$

LaTeX

Docker

Blender

Google Colab

AWS

Experience

Computer Vision Engineer Co-op

3DM Devices

May 2023 – Dec 2023 Aldergrove, British Columbia

Computer Vision for industrial inspection on low contrast heightmap images of tires
Detected barcodes on tires with an mAP of 92.8 using FasterRCNN object detector that performs 6x faster than an image processing-based detector
Segmented defects on tire sidewalls with an mIoU of 72.8 using DeepLabv3+ for multi-class semantic segmentation
Identified distorted tire beads using a rule-based system developed by feature engineering tabular data from tire scan metrics
Curated and annotated machine learning datasets from raw tire scans for object detection and semantic segmentation

Data Scientist

Eltropy

Nov 2021 – Jul 2022 Bangalore

Forecasted text messages for Credit Unions in the US categorized by departments and message types by implementing an ML pipeline using AWS cloud services (Amazon Forecast, Amazon MWAA)
Showcased KPIs for Credit Unions by designing performance optimized dashboards on AWS QuickSight and embedded the dashboards on the Eltropy application to be sold as premium insights to customers

NLP Engineer

LimeChat

Jan 2021 – May 2021 Bangalore

Boosted sales by 6% for The Man Company - a leading D2C men’s grooming brand by designing and implementing a system to launch personalized chat re-marketing campaigns.
Improved the chatbot’s ability to handle typos and synonyms for the product discovery quiz by creating a Parts of Speech (POS) Tagging based system to obtain synonyms and typos from real user conversations with the chatbot.

NLP Research Assistant

CVIT, IIIT Hyderabad

Jul 2020 – Dec 2020 Hyderabad

Achieved state-of-the-art BLEU scores 19.07, 19.18 and 9.48 for English to Tamil, Urdu and Odia translation respectively and WAT20 En-Odia leaderboards rank 3 by fine-tuning a multilingual neural machine translation model.

Computer Vision Research Intern

IMAGINE Team, LIRIS Lab, Ecole Centrale de Lyon

May 2019 – Jul 2019 Lyon, France

Increased mIoU score to 46.65 for semantic segmentation of fruits by creating a synthetic training dataset of fruit tree images using Blender and enhancing its photo-realism using generative adversarial networks (CycleGAN).

Computer Vision Research Intern

IIT Bombay

Dec 2018 – Apr 2019 Mumbai

Created and utilized a labelled image dataset of people smoking cigarettes to finetune an Inception-ResNet-v2 model to classify images containing smokers, achieving an accuracy of 74.66% and F1 score of 0.7055.