Professional Experience

Joining a high-output 4-member engineering team at Jawaharlal Nehru Technological University, I contributed to a complete overhaul of the university's web infrastructure. We modernized legacy portals into responsive Next.js applications backed by Drizzle ORM and PostgreSQL. To transform how students and faculty interact with university information, I developed an intelligent RAG campus assistant that instantly answers queries about official notices, admissions, examination schedules, and academic regulations in natural language.

My work at UptoSkills centered on bridging statistical machine learning models with live production web services. I built predictive and classification pipelines utilizing TensorFlow and Scikit-learn, optimizing data preprocessing and feature extraction to handle noisy datasets. To make model outputs accessible to downstream frontend platforms, I wrapped the machine learning logic into lightweight, high-performance Flask REST API endpoints capable of real-time inference.

During this internship focused on sustainable technology solutions, I built deep learning computer vision models to classify electronic waste with 92% accuracy, aiding automated recycling sorting. Additionally, I conducted predictive modeling on a 50,000+ record dataset to forecast electric vehicle (EV) adoption trends with 87% accuracy. By applying post-training quantization and model pruning, I streamlined deep learning model weights, reducing deployment overhead by 40% and drastically boosting inference speed.

At NTS Nihon Global, I focused on high-concurrency backend architecture and legacy codebase modernization. I architected 12+ robust PostgreSQL-backed REST APIs designed to support more than 10,000 daily active users securely and efficiently. By diagnosing performance bottlenecks, refactoring database queries, and implementing backend caching strategies, I helped reduce application load times by 45%, driving a 28% increase in overall user session retention.

Starting my internship journey at YoungBot Academy, I automated large-scale data ingestion by building robust web scraping and ETL pipelines using Scrapy and Pandas, processing over 100,000 raw data records effortlessly. On the frontend, I modularized the React codebase using reusable component architecture, dynamic code splitting, and lazy loading, cutting initial page bundle sizes and accelerating load times by 55%.