Sarvesh Siras

About me

I am a graduate student at Carnegie Mellon University with experience as a Software Engineer II at Mastercard. I have a strong interest in machine learning and problem-solving. I am passionate about learning new things and exploring innovative ideas. I enjoy reading Technology blogs and watching Tv-series.

Education

Master of Information Systems Management

Carnegie Mellon University, Pittsburgh, PA

Aug 2024 - Aug 2025

B.Tech in Information Technology

(With Honors in Computer Engineering)

College of Engineering Pune (COEP), Pune, Maharashtra

Aug 2017 - May 2021

Research Work

A Deep Learning Approach for Digitization of Invoices

Skills

Programming Languages

C++

Python

Java

MySql

HTML

CSS

JavaScript

PHP

Platforms and Frameworks

My Experience

Professional

Software Engineer II

July, 2021 - July, 2024

Mastercard, Pune, India

Designed and Developed highly scalable, secure, fault-tolerant micro-services capable of handling 500 transactions per second using Java, Spring boot, Kafka, and SQL
Played a vital role in the migration of Monolithic applications to Micro-services architecture which had better performance and scalability
Utilized tools like Splunk and Dynatrace to analyze the performance of critical features, identifying and resolving bottlenecks to enhance overall system efficiency threefold
Collaborated effectively with stake-holders, product managers, architects, and fellow engineers to define project requirements, plan Agile development sprints, and deliver high-quality software solutions on time
Thrived in a fast-paced environment, effectively managing multiple projects and priorities, actively participating in code reviews and provided constructive feedback to enhance code quality

Summer Intern

May, 2020 - June, 2020

Mastercard, Pune, India

Worked with Operations and Technology team to develop a REST application using technologies like Spring-boot, Java, React, and H2 Database
Developed a mocking framework for enabling Rapid Application Development that reduced development time by 50%

Leadership

Pro Bono Consultant

August, 2021 - October, 2021

Bal Utsav, India

As a Team Lead, managed a team of eight over the 10-week program to develop social impact recommendations and solutions to support Bal Utsav in creating a communication strategy with a strong focus on digital visibility to build awareness and engagement around its programs and services to further expand its reach

Head, Bhau's Innovation and Entrepreneurship Cell

College of Engineering Pune Technological University, Pune, India

Responsible for promoting spirit of innovation and entrepreneurship among my fellow students in Institution
Played a crucial role in the organization and scheduling of Pune Startup Fest, which witnessed a convergence of 200 startup exhibitors, 100+ investors, and a crowd turnover of 20k+ people

Projects

My Projects

RegHealth Navigator

Developed an AI-powered compliance assistant for Medicare policy analysis using RAG architecture with OpenAI’s GPT-4o-mini and FAISS-based vector search
Built and maintained Flask-based backend APIs for Q&A, summarization, and document comparison modules integrated with a React + TypeScript frontend
Implemented citation-grounded Q&A and hybrid retrieval (semantic + keyword) to ensure high accuracy and traceability in regulatory responses
Achieved 53% reduction in time and cost for regulation analysis, with 95% accuracy and strong validation through LLM-based evaluation and user testing

Digitization of Invoices

Developed a system to digitize paper-based invoices with tables into excel sheets using Computer Vision techniques
The system used Machine learning models to map the invoice data into the excel sheet
The system used the Convolutional Neural Network model to detect table and Tesseract and Image processing using OpenCV to process text
The system had an accuracy of around 86%

File Transfer Protocol

Developed a secure terminal-based bidirectional File Transfer Protocol application for transmitting files over the network using a TCP/IP connection
The application allowed the transfer of multiple files simultaneously with the help of threading
The application was developed in Python using Socket Programming libraries

Production-Ready Movie Recommendation System

Designed a movie recommendation system using collaborative filtering, serving 1M+ users with Python/Flask, achieving 99% service availability, handling >2K daily requests with <800ms latency
Designed modular ML pipeline with Kafka-based ingestion, feature engineering, and model training using Python
Automated infrastructure testing and CI/CD with Jenkins and pytest, achieving 90%+ code coverage
Containerized services with Docker and implemented rolling updates, Prometheus-Grafana monitoring, and A/B testing infrastructure and load balancing using nginx
Analyzed fairness, security, and feedback loops using telemetry data; detected bias and proposed mitigation for data poisoning attacks

Determining the Popularity of Politicians from Tweets

The project's objective was to determine a politician's popularity using tweets mentioning them during election campaigning
The system gathered tweets using Twitter API, and these tweets were then tokenized to keep only the relevant data
Finally, the tweets were analyzed using Sentiment Analysis techniques and Machine learning models to determine the popularity of the politician

Heap Manager

Developed a set of C language functions like malloc, calloc, realloc and free, to manage the system heap memory
Used a combination of buddy and free-list type memory allocators to allocate and manage the system heap
The functions were written in C language and used brk and sbrk system calls for memory management