Ram Srivastava

StudyDeck – Academic Productivity Ecosystem (Feb 2024 – Apr 2025)

• Architected a multi-platform productivity hub designed for centralized academic resource distribution and automated schedule management.
• Engineered a high-performance frontend using Next.js and SCSS, leveraging Zustand for streamlined state management and Zod for rigorous data validation.
• Optimized server-side delivery via Nginx to manage a repository of 400+ GB of scholarly data, incorporating a custom GPA calculation engine to enhance user utility.

BOSM-Team Registration Interface (Jun 2024 – Jul 2025)

• Developed a specialized registration portal for the BITS Open Sports Meet (BOSM), specifically tailored to manage team-based entries from 20+ external institutes.
• Implemented secure authentication and team-management workflows using Firebase and JWT, while integrating Stripe to handle high-volume financial transactions.
• Facilitated the onboarding of 500+ participants across 15+ diverse sports categories, maintaining strict data privacy through robust encryption protocols.

Automated Fuel Retail Ledger & Workforce Management (Oct 2025 – Nov 2025)

• Engineered an automated financial auditing system to generate real-time balance sheets for fuel station operations, streamlining the tracking of daily sales, inventory levels, and cash flow via GO/SQL.
• Architected a smart payroll module that synchronizes staff attendance records with salary disbursement, automatically detecting absences and calculating precise deductions to ensure 100% payroll accuracy.
• Optimized a centralized database to handle daily transactional logs and employee records, reducing manual administrative overhead by 35% and eliminating human error in financial reporting.

Violence Detection in Video Content (Dec 2025 - Jan 2026)

• Engineered a robust Deep Learning system to identify and categorize violent sequences within video streams, automatically generating real-time trigger warnings to safeguard user mental health.
• Developed a frame-processing pipeline utilizing OpenCV to convert video data into image frames for training Convolutional Neural Networks (CNN), leveraging pre-trained architectures like ResNet50 and VGG19 for optimized feature extraction.
• Classified a dataset of 2,000+ video samples, implementing high-accuracy detection protocols to filter sensitive content and mitigate the impact of harmful visuals on digital platform.