Ashutosh Sharma Email & Phone Number

Cambridge, Ma, Us

Urbana-Champaign Area

CS 441 - Applied Machine Learning, Prof. Derek Hoiem, Department of Computer Science, UIUC

Seattle, Washington, United States

Finetuning LLMs for Entity Matching.

Urbana-Champaign Area

CS 173 - Discrete Structures, Prof. Benjamin Cosman, Department of Computer Science, UIUC

Mumbai, Maharashtra, India

Under the guidance of Prof. Pushpak BhattacharyyaProject 1 - Neural Information Retrieval for Enterprise Search System• Implemented ColBERT model with query expansion using Dense Pseudo Relevance Feedback• Modified Attention Mechanism in transformer architecture using Dependency Parsing to encode syntaxinformation and achieved 47% improvement in MAP@10 and MRR@10 over ColBERTv2-based strong baselineProject 2 - Multilingual Information Retrieval for Zero-shot Question-Answering• Curated IR dataset IndicMARCO for 11 Indian languages by translating MSMARCO dataset using NLLB model• Developed IndicColBERT, a multilingual neural information retrieval model using mBERT encoder• Integrating IndicColBERT retriever in Retrieval Augmented Generation Models for question-answering

Mumbai, Maharashtra, India

Mumbai, Maharashtra, India

Image Translation for real-time CT Scan generation from UltrasoundsGuide: Prof. Amit Sethi, Department of Electrical Engineering, IIT Bombay• Utilized Fourier Domain Adaptation for transforming simulated ultrasounds to real ultrasound images• Deployed pix2pix and CycleGAN Models in PyTorch to learn the mapping from ultrasounds to CT Scans• Achieved 0.87 SSIM and 0.008 MSE, validating the efficacy of using simulated data for medical image translation

Bengaluru, Karnataka, India

Developed a Trading Platform in Flask & ReactJS; received Pre-Placement Offer for exceptional performance• Integrated Apache Kafka with MongoDB for persistent storage of Stock Prices from external APIs• Developed a Trade Matching Engine for efficiently queuing and settling trades & an interactive GUI• Encrypted data in transit and performed unit testing in Pytest to ensure High Code Coverage

Mumbai, Maharashtra, India

Developed an E-Commerce Webpage Classification System using NLP and Deep Learning• Reviewed literature on employing SVD, RNNs, CNNs and N-Gram Models for webpage classification• Developed data pre-processor and an RNN Model in Keras python framework and integrated TransferLearning by using pre-trained Glove vectors and achieved 86% Accuracy on testing dataset

Mumbai, Maharashtra, India

London, England, United Kingdom

o Studied the high quality statistics on Europe in Eurostat online database and ideated efficient data processing stepso Automated creation of NUTS-level dictionary in Europe from Eurostat online database using Pandaso Prepared dataset using MS Excel from Genworth website for predicting lucrative real-estate locations

Mumbai, Maharashtra, India

MA 108 - Differential Equations, Prof. U. K. Anandavardhanan, Prof. Prachi Mahajan, Department of Mathematics, IIT Bombay

Mumbai, Maharashtra, India

Mumbai, Maharashtra, India

Master Of Science - Ms, Computer Science, 4/4

Courses: Theoretical Reinforcement Learning Advanced Topics in Security, Privacy, and Machine Learning Secure Multiparty Computation.

Bachelor Of Technology - Btech, Mechanical Engineering, 9.1/10

Honors (Thesis: Improving Textual Information Retrieval) and Dual Minors - Computer Science & Machine Learning Relevant Courses: 1).