Shubham Bhadra

Teaching Assistant for DADS7275 — Machine Learning and Data Analytics under Prof Dinesh - Lead TA team - Evaluated 100+ weekly submissions for a cohort of 110 students, grading notebooks, projects, and labs using Canvas rubrics to ensure consistent and timely feedback. - Resolved 50+ student support tickets via Help Scout, addressing submission issues, grading discrepancies, and technical challenges. - Provided academic support through weekly office hours, clarifying assignment concepts, reducing recurring errors, and improving overall submission quality.

Teaching Assistant for INFO6150 — Web Design & User Experience under Brett Ritter • Assisted in teaching web development concepts including HTML, CSS, JavaScript, and React to 30 students. • Graded assignments and provided feedback to ensure clarity and improvement in coding practices. • Guided students in debugging, responsive design, and UI/UX best practices through hands-on support. • Managed the class independently, addressing technical doubts and maintaining effective communication with the instructor. • Conducted doubt-solving sessions to reinforce concepts from lectures and assignments.

DigitalOcean Marketplace team • Contributed to DigitalOcean, a leading cloud provider for developers and startups, specifically improving the DigitalOcean Marketplace, which offers pre-configured apps and services deployable on cloud servers. • Built production-grade LLM Inference Service powering DigitalOcean Marketplace Search in Go (gRPC, Protobuf, Redis, RAG), reducing p90 query latency from ~3.2s to 1.5s over 8 weeks and improving reliability across thousands of queries. • Designed evaluation pipelines with a golden dataset, increasing click-through rates by 15% and deployments from search by 10% (measured over the internship period). • Improved internal RPCs and enhanced the Vendor Portal & Pre-deploy service to streamline Marketplace partner onboarding and integration, collaborating in Agile sprints. • Created and deployed custom Marketplace Droplets with Shell scripting and Packer, automating provisioning and cutting deployment setup time for end users from hours to minutes. • Implemented monitoring and automated testing pipelines for Marketplace Droplets, significantly reducing release-related failures and ensuring reliable, consistent deployments.

• Scaled PayZapp, HDFC Bank’s digital payments and mobile wallet platform, by optimizing RESTful APIs (Swagger/OpenAPI, Node.js microservices), handling 35% higher transaction throughput post-optimization. • Architected a real-time chat system (microservices + WebSocket), rolled out to thousands of concurrent PayZapp users. • Containerized services with Docker, deployed via Kubernetes, and implemented CI/CD pipelines to ensure seamless, efficient releases and reduce deployment time by 25%. • Collaborated across teams to identify performance bottlenecks, boosting system responsiveness and improving user engagement across millions of active users.

• Conducted EEG signal research to detect seizures, building Python pipelines for real-time high-frequency time-series data processing using wavelet transform and CNN-LSTM • Automated Wavelet Transform for data processing and implemented multi-threaded transformation to improve efficiency • Improved machine learning classifier accuracy to 85% by preprocessing EEG signals with bandpass filtering, ICA, and advanced feature extraction

Data : • Developed a data pipeline using OpenPyXL and PyPDF2 to extract structured data from hospital reports into Excel, reducing manual entry time by 30% and saving 10 hours weekly • Developed a ML-based model achieving 87.5% accuracy in detecting Covid-19 patients, optimizing data preprocessing and visualization techniques to improve patient care and resource allocation Software : • Designed and implemented University’s Event Management System with a React.js front-end and Node.js back-end • Built RESTful APIs to enable seamless data integration with a MySQL database, managing registrations and scheduling for over 500 participants per event • Created dynamic user dashboards for participants and admin panels for event management tasks. • Enhanced application performance and scalability through microservices architecture, increasing system capacity to handle a 30% growth in event registrations and database operations efficiently

Object Oriented Design - Java Distributed Systems Web Development Algorithms Data Science Design Patterns - Java Cloud Computing Enterprise Software Design - Java