experience.
Doctoral Researcher — The Research Foundation for SUNY
May 2025 - Present · Buffalo, NY
Graduate Research Assistant — University at Buffalo
January 2024 - May 2025 · Buffalo, NY
- Guided by Dr. Karthik Dantu
- Researched plastic-type classification on the US Plastics and MRF dataset, employing classification models such as InceptionNet, ResNet, and Vision Transformer, as well as instance segmentation models like Mask-RCNN, YOLO.
- Performed activation & ablation analysis of the outputs using Grad-CAMs, Eigen-CAMs, and Ablation-CAMs
- Extracted diffuse reflection from images using Autoencoder/VAE-based decomposition and applied the cleaned diffuse output to achieve more reliable image classification.
Graduate Research Assistant — University at Buffalo
January 2024 - May 2025 · Buffalo, New York, United States
- Guided by Dr. Rohini Srihari.
- Enhanced the Alternative Augmentative Communication (AAC) using the RAG framework.
- Built Real-Time Data Retrieval Module and Context Generation using LLMs and fine-tuning them
Graduate Teaching Assistant — University at Buffalo
May 2023 - May 2024 · Buffalo, New York, United States
- Under Dr. Detrich Khulmann and Dr. Jonathan Lopez
- Courses Assisted: Probability Theory, Introduction to Time Series in R, Analysis of Experimental Data
- Guided 200+ students across 3 Semesters
Applied Scientist — Rambus
July 2024 - October 2024 · San Jose, California, United States
- Finetuned Models like BART for Query Correction, T5 for Query Expansion and Llama2-7B and Mistral-7B using LORA.
- Implemented Monitoring of a compound AI system using UniEVAL, RADE, and FEQA.
- Developed a comprehensive RAG (Retrieval-Augmented Generation) system to facilitate efficient document search and precise answer extraction tailored for the Semiconductor industry.
- Implemented a multi-component architecture incorporating BERT-based Classifiers from Hugging Face and fine-tuned them for initial document filtering, NER (Named Entity Recognition) models like for entity extraction, and Vector Databases for semantic search. Enhanced the system's capabilities using RAG powered by OpenAI and Flan-T5 models from Hugging Face for contextual answer generation.
- Used AWS Comprehend for text analysis and NER.
- Ranking Documents using Retrieval Bi-Encoder and Re-Rank Cross-Encoder from Sentence Transformer from Hugging Face.
- Leveraging AWS Textract for document analysis in RAG system
- Deployed model in AWS Sagemaker and AWS EC2 instances.
- Amazon SageMaker Model Monitor to track and analyze deployed models in production environments.
- And used S3 to store the models using IAM roles.
Data Scientist — Thoucentric
February 2021 - December 2022 · Bengaluru, Karnataka, India
- Conducted sentiment analysis on reviews within the Research and Development department of Unilever, utilizing Recurrent Neural Networks (RNNs), BERT-based classifiers from Hugging Face and fined tuned them. Identified areas of improvement within the company's products based on distinguished sentiment, providing valuable insights for improving product offerings. Increased annual product sales from 54% to 56.16%.
- Created an Information Retrieval system (IR) to accurately identify and map products across various sources and marketplaces, facilitating efficient mapping and integration of products between different platforms and marketplaces for the company's legacy data. Facilitated initiation of an internal project valued at $10,000 for the company.
- Amazon SageMaker Model Monitor to track and analyze deployed models in production environments.
- And used S3 to store the models using IAM roles.
Cloud Infrastructure Engineer (Data Science Solutions) — CloudOptimize from CloudAccel Inc.
March 2018 - February 2021 · Hyderabad, Telangana, India
- Integrated Auto Encoders to detect machine anomalies, increasing monthly alerts from 300 to 450, and enhancing monitoring and detection capabilities with optimized data and deployed model using AWS EC2 Instances.
- Migrated client's infrastructure from on-premises to AWS, saving $60,000 annually, and transferred a 15GB database from AWS to Azure with only 2 minutes of downtime.
- Designed scalable cloud environments for high availability and performance, completed a $5,000 project, and implemented infrastructure as code, reducing deployment time from 5 hours to 0.5 hours.
- Using AWS Autoscaling, AWS Load balancer, and IAM permissions.
- Developed a comprehensive monitoring system with ELK Stack, improving issue identification by 50% and ensuring timely resolution, reducing downtime and maximizing operational efficiency.
- Built and deployed automation bots with Blue Prism and UI Path, reducing manual workload by 10 hours per week through streamlined email management and automated file downloads.
AI Researcher — Sri Venkateswara University
June 2017 - March 2018 · Tirupati, Andhra Pradesh, India
- Building Machine Learning Models for predicting if an auto insurance policy holder files a claim using XGBoost, Logistic Regression (Ensemble), Support Vector Machines and Deep Neural Networks.
- Removed imbalance in the data set using SMOTE.
- Analyzed different CNN architectures for classification using GRAD-CAMS.
- Deployed models on EC2 instances using FLASK.
education.
University at Buffalo
Doctor of Philosophy - PhD, Computer Science and Engineering · August 2025 - July 2029
University at Buffalo
Master's degree, Data Science · January 2023 - May 2024
Sri Venkateswara University
Bachelor of Technology, Computer Science and Engineering · October 2013 - May 2017