Pursuing my Master's at the University at Buffalo was a passion-fueled journey that solidified my love for computer science. The program's challenging curriculum pushed me to move beyond foundational concepts and truly master the intricacies of cutting-edge fields like Deep Learning and Algorithm Design. I particularly valued the hands-on projects that allowed me to build, test, and refine my ideas, turning theoretical knowledge into tangible, impactful solutions. This experience has been invaluable in shaping my perspective as an engineer.

Courses:

• Deep Learning
• Data Modelling and Query Language
• Algorithm Analysis and Design
• Intro to Machine Learning
• Data Intensive Computing
• Operating Systems
• Computer Security
• Digital Product Management
• Computer Vision and Image Processing

Overall GPA during Masters = 3.96 / 4

Developed a basic understanding of Concepts in Software, Web Scripting Technologies, Database Management, Algorithms and Software Design

Overall CGPA during Bachelors = 9.36 / 10 => GPA = 3.74 / 4

Technologies: Java | SpringBoot | Kafka | RabbitMQ | Camunda Modelling | Elasticsearch | AWS | Angular | Node.js | Nest.js | Drools | JUnit | Mockito

• Led the end-to-end architecture and scalability of 8+ core microservices using Java, Spring Boot, Node.js, and Kafka, designing robust RESTful services for seamless integration with financial systems and implementing CI/CD pipelines.
• Spearheaded the design of secure BKYC and Wallet Transfer modules, ensuring 99.98% adherence to global AML/CTF regulations while handling sensitive financial data, reducing fraud by 40%.
• Promoted to Lead BPM Design for KYC case workflow after re-architecting a three-tier Camunda workflow engine, which slashed SLA breach rates from 42% to 8%, addressing the bottlenecks in the BKYC lifecycle.
• Engineered a custom OCR and Fuzzy Matching pipeline to extract and validate document IDs against internal records with an 80% similarity threshold, automating approval workflows and reducing manual review overhead.
• Authored API and user documentation and deployment guides, mentored 3 new developers to facilitate rapid onboarding and drive best practices in code reviews.
• Managed project releases to production, adhering to SDLC best practices, with less than 1% post-release defects, sustaining 99.9% system uptime through active monitoring with Elasticsearch.
• Optimized CouchDB database queries, resulting in 33% reduction in query latency and improved application performance.

• Implemented a Grievance Redressal System using the MERN stack, enabling time-bound escalation of complaints across management levels with a timeline-style UI for real-time tracking, improving employee satisfaction by 53%.
• Established secure CRUD operations for database interactions featuring AES encryption protocols to ensure the protection of ticket data across applications, resulting in streamlined processes handling upwards of 5K employee submissions monthly.

Successfully earned the AWS Certified Developer - Associate certification, demonstrating expertise in developing, deploying, and debugging cloud-based applications using AWS services.

Key Skills Validated:

• Development and deployment of cloud-native applications on AWS
• Proficiency in AWS SDKs, APIs, and CLI tools
• Implementation of CI/CD pipelines using AWS services
• Application of AWS security best practices and IAM
• Working with AWS Lambda, API Gateway, DynamoDB, and S3
• Debugging and monitoring applications using CloudWatch
• Understanding of serverless architecture patterns

Published research paper at the IEEE CogMI 2025 conference on "Self-Supervised Learning for Image Representation Learning and Classification of Alzheimer Disease".

Research Focus:

• Self-Supervised Learning techniques for medical image analysis
• Image representation learning for Alzheimer's Disease classification
• Deep learning applications in healthcare and medical imaging
• Computer vision and pattern recognition in medical diagnostics

This research contributes to the field of medical AI by exploring advanced self-supervised learning methods for improved Alzheimer's Disease detection and classification from medical images.

Technologies: Object Detection | Object Classification | YOLOv3 | Logistic Regression | Custom Clock Algorithm | Python

• Formulated a model to predict the traffic signal wait time based on real-time traffic analysis using Object detection using YOLO v3 single-stage predictor.
• Model able to predict duration accurately up to 87.68% with reducing vehicle wait time by 8 minutes for daily commuters.
• Performed Analyses of the collected data to draw patterns of traffic, driving style, need for public transportation and designed the dashboard

Detailed Project

• Using the surveillance camera at traffic signals, we will capture the images of incoming roads to signal.
• Using Yolo V3 Model to detect vehicles into 4 category- Bus, Truck, Car, Motorcycle with the count of each present
• Using this as a input to our trained model to estimate the time required to optimally free the traffic.
• Model was developed based on data of time taken by vehicle to clear in traffic signal once it is green, vehicle 0-30 acceleration, Size of road occupied by vehicle, etc.
• Based on input parameters we developed a Neural Network model to accurately predict the signal time
• Integrating the signals and adjusting times to ensure minimal wait times at signals
• This would reduce wait times for route with heavy traffic which would wait at cross section where not many vehicles are present
• Dynamically updating signal times helps to streamline the traffic without the need to wait at signals, reducing travel time

Technologies: PHP | JavaScript | MySQL | HTML | CSS

A comprehensive web application designed to track and manage student activities beyond academics, providing a centralized platform for students, teachers, and staff at PICT, Pune.

Key Features

• Database Design: Designed and implemented a robust database to keep track of all student activities apart from their academics.
• Dashboard & Analytics: Provided a comprehensive dashboard to visualize statistics, generate reports, and display personalized profile pages for each student.
• Role-Based Access Control (RBAC): Implemented an RBAC model for approval/permissions for any event/activity, ensuring proper access control and workflow management.
• Reward System: Added a reward module for all-round performers, including badges and other achievements, linked to their profile page to share with colleagues.
• Real-World Impact: The system is currently being used within PICT, Pune, helping students, teachers, and staff to efficiently keep track of activities and student engagement.

Technologies: PHP | Angular | MySQL | Razorpay

Led backend development for the Annual Tech Fest Event PICT INC 2020, creating a comprehensive website with online event registration and payment integration.

Key Contributions

• Backend Leadership: Served as Lead Backend Developer for the Annual Tech Fest Event PICT INC for year 2020.
• Online Registration System: Updated the website to add online event registrations, streamlining the registration process for participants.
• Payment Integration: Integrated Razorpay online payment platform, enabling secure and seamless payment processing for event registrations.
• Full-Stack Development: Worked with PHP for backend, Angular for frontend, and MySQL for database management, ensuring a robust and scalable solution.

Built an end-to-end full-stack application during UB Hacking designed to automate the tedious process of tailoring resumes to specific job descriptions. Instead of basic templates, ResuMatch intelligently selects and rewrites user experiences to match target roles.

Core Technologies: React, Tailwind CSS, FastAPI, MongoDB, Sentence Transformers, DeepSeek API

Key Features & Technical Implementation

• Full-Stack Architecture: Developed a responsive frontend using React and Tailwind CSS to collect user data, storing master profiles in MongoDB. Engineered a FastAPI backend to seamlessly orchestrate the database and AI processing layers.
• Semantic Matching Engine: Implemented Sentence Transformers to analyze the true essence of a Job Description (JD). The system calculates semantic similarity against the user's master profile to automatically rank and extract the top 3 most relevant projects and top 2 work experiences.
• AI-Driven Generation: Integrated the DeepSeek API to process the filtered experiences, automatically summarizing and rewriting raw data into strong, professionally framed bullet points aligned perfectly with the JD.
• System Integration: Successfully tackled complex debugging and data-flow challenges to bridge the gap between the frontend UI, MongoDB, the semantic ranking model, and the generative AI layer within a tight hackathon timeline.