Akash Kumar

Application Developer

Software Developer

Python & SQL Developer

Cross-Platform App Creator

React Native Enthusiast

Projects

Dialogue Director (FYP)

10/2024 – Present

  • Achieved A grade and ranked 3rd best project among 50+ in the final year showcase.
  • Developed a Python backend for YouTube subtitle summarization, reducing manual review time by 65%, and integrated it with a React Native app for real-time video analysis and results.
  • Integrated APIs for speaker diarization and topic segmentation, improving transcription clarity by 40%, while ensuring seamless communication between the frontend and backend.
  • Processed 50+ videos, demonstrating scalability, efficient topic indexing, and preparing for future sentiment analysis integration to expand use cases by 30%.

ClearSight – AI Powered Vision for the Blind

  • Developed an assistive mobile app for visually impaired people using React Native (frontend) and Flask (backend).
  • Implemented real-time object detection (YOLOv8), color recognition (OpenCV), and OCR-based text-to-speech for reading printed content.
  • Integrated Whisper (OpenAI) for voice commands, enabling hands-free navigation.
  • Built an emergency assistance system using Firebase + Twilio, allowing users to send instant alerts to pre-saved contacts.
  • Designed for affordability and accessibility, ensuring smooth performance on low-cost Android devices and compliance with WCAG standards.

Patient Medicine Alarm App

  • Created a mobile reminder app that increased medicine adherence in test users by 25% using timely alerts.
  • Achieved 95% alert accuracy during testing on multiple Android devices.
  • Reduced scheduling errors by 40% with built-in time validation and persistent alerts.

Bank Management System (C#)

  • Built a banking app that handled over 200+ user accounts, enabling secure deposits, withdrawals, and reports.
  • Reduced transaction processing time by 30% using optimized code and real-time updates.
  • Implemented error handling to bring input error rate down by 80% during simulations.

Client-Server Chat Application (MPI-Based)

  • Engineered a chat system with Message Passing Interface (MPI) for 100+ users, maintaining average latency under 50ms.
  • Reduced communication overhead by 45% through efficient socket design and concurrency handling.
  • Successfully simulated multi-client load testing across different machines using distributed architecture.