Research Flow

An intelligent AI agent designed for company research and account planning.

Autonomous Research : Automatically gathers and synthesizes data from the web using smart agents.
Dynamic Account Plans : Generates and updates structured plans in real-time based on your conversation.
Visual Data : Creates instant flowcharts and diagrams (Mermaid.js) to visualize complex information.
Context-Aware RAG : Utilizes ChromaDB and vector search for deep document analysis and retrieval.
Voice Enabled : Supports natural voice interactions with real-time text-to-speech playback.
Modern Tech Stack : Powered by FastAPI, LangGraph, React, and TypeScript for high performance.
File Intelligence : Upload and analyze PDFs or docs seamlessly within the chat interface.
Persistent Sessions : Keeps track of your research history and planning progress automatically.
Interactive UI : A clean, responsive interface built for efficient professional workflows.

No More Brainrot

A browser extension for Firefox that blocks digital distractions like YouTube Shorts and Instagram Reels.
Utilizes wildcard pattern matching for custom site blocking.
Redirects blocked content to user-selected sites for productive browsing.
Tracks number of blocked attempts and provides real-time statistics.
Pause functionality allows timed disabling, optionally secured with a passphrase.
Local storage only—no cloud sync or data tracking.
Modern popup interface with dark theme.
Session-based authentication for privacy.
Lightweight and efficient; runs only during blocking events.
Ideal for students, professionals, and families seeking digital wellness.
Open source, easily auditable and modifiable.
Roadmap includes time-based blocking, creator whitelisting, and Chrome/Edge support.

pyCHIP8-neo

A CHIP-8 emulator in Python and PyQt6, simulating classic 8-bit hardware.
Supports all CHIP-8 instructions, registers, timers, and RAM.
Interactive GUI for running, pausing, and step-by-step debugging of classic games.
Graphical 64x32 monochrome display and virtual hexadecimal keypad.
Built-in ROM library (Tetris, Pong, etc.) and easy ROM management.
Modular code divides core emulation, GUI tools, and data handling.
Unique for educational purposes: teaches VM architecture and emulation basics.
Debugging panels allow inspection of CPU state, stack, and display memory.
Cross-platform: compatible with Windows, Mac, Linux.
Efficient multithreading for concurrent CPU and timer execution.
Extensible and open source for hobbyist modification or variant support.

QuizBuzz

A retro-styled web quiz platform for education and self-learning.
Supports interactive quizzes with multi-select questions and LaTeX math rendering.
JSON-based data format with Pastebin and direct URL integration.
Built-in Data Builder to merge quizzes, Slide Builder for PowerPoint export, Validator for syntax checks.
Accessible via desktop, tablet, or mobile browser.
Personalized quiz settings with local preference storage.
Theme customization and responsive design.
Real-time progress tracking and audio feedback for user engagement.
Ideal for classroom, training, or solo practice sessions.
Easy hosting on GitHub Pages or a local web server.
Extensible, open source, and optimized for accessibility.

WordleAI

A Python-based entropy solver for Wordle-type word puzzles.
Uses information theory to select optimal guesses and solve puzzles quickly.
Supports 4–12 letter words with dynamic database management.
Analyzes feedback to eliminate impossible solutions after each guess.
Computes entropy for all options to maximize information gain.
CLI and FastAPI web frontend for interactive probability rankings.
Keeps history of used guesses and reranks choices in real time.
Efficient filtering and session management.
Open source, ideal for research in algorithmic puzzle solving and linguistics.

SafeJourney-AI

A driver drowsiness detection system using machine learning and computer vision.
Processes live camera feeds to track facial landmarks—eye and mouth aspect ratios.
Flags drowsiness when behavioral metrics cross defined thresholds.
Live dashboard built with Dash/Plotly visualizes real-time data and alerts.
Triggers audio warnings for safety when fatigue is detected.
Optimized for high-resolution cameras and cross-platform operation.
Modular codebase: facial analysis, prediction, alerting, and visualization modules.
No network needed—full privacy and local processing.
Designed for use in vehicles, with future support for more biometric sensors.

Winner - Smart India Hackathon 2024

Our team, RisingPhoenix076 , emerged as winners of Smart India Hackathon 2024. We addressed the problem statement "Video call intercom based on analog/IP system with vibration sensor" (PS-1578), proposed by the Indian Sign Language Research And Training Centre (ISLRTC) under the Ministry of Social Justice & Empowerment.

This project holds a special place in our hearts as it is designed to empower individuals with hearing disabilities, making communication more accessible.

Our Solution

Our solution included: - Two devices enabling internet-free video calls with conferencing support. - Custom features like text-to-speech, Automatic Speech Recognition (ASR), and unique lighting notifications. - A dedicated SOS mode for emergencies. - A wearable device for accepting and receiving calls remotely.

The jury from ISLRTC was incredibly supportive and encouraged us to develop this project into a market-ready product.

Team & Mentor

• Teammates : Kiruthika S, Saai Srivathsan, Kishore .K, Jayachandiran Kumar, Mugilan Anbarasu
• Mentor : Dr. Priya E

11th Place - Amazon ML Challenge 2024

As part of team Miracle Workers , we participated in the Amazon ML Challenge 2024, securing 11th place nationally among top institutions like IITs, NITs, and IIITs. We achieved an F1 score of 0.715.

The task involved identifying specific product parameters (e.g., weight, volume) from a massive dataset of 2.6 lakh training images and 1.3 lakh test images.

Key Challenges Overcome

• Handling Large Datasets : We dynamically loaded data from AWS to manage the nearly 4 lakh images efficiently.
• Computational Resource Shortages : We utilized a Tree of Thoughts (ToT) approach and set up a local server to coordinate multiple Colab and Kaggle instances, enabling distributed computing with T4 and P100 GPUs.
• Data Inconsistencies : We wrote regex scripts to standardize unit representations (e.g., 'cm', 'centimeters') and handle fractional units.

Team

• Kiruthika S
• Saai Srivathsan
• Kishore .K

Research Paper: Water Quality Modeling

I am a co-author of the published research paper, " Water Quality Parameters Modeling of Thamirabarani River ." This study analyzes the physical, chemical, and biological indicators of the river's water quality using advanced machine learning algorithms.

Methodology & Findings

We utilized data from the Tamil Nadu Pollution Control Board and applied several machine learning algorithms, including: - Multiple-Linear Regression - Polynomial Regression - Support Vector Machine (SVM) - k-Nearest Neighbors (kNN) - Random Forest

The Random Forest algorithm proved to be the most effective, achieving an accuracy of 96% , offering valuable insights for policymakers and environmentalists.

Authors & Mentor

• Authors : Dinesh (myself), Saai Srivathsan, Sachin M P
• Mentor : Dr. K. Sri Dhivya Krishnan

IEEE "Best Student Volunteer Award 2023"

I was honored to receive the prestigious IEEE "Best Student Volunteer Award 2023" . This award recognizes my dedication and contributions to the IEEE community.

Leadership & Impact

As the Chairperson of the IEEE Engineering Medicine and Biology Society (EMB) Sairam , I am proud of the work our team has accomplished. We successfully organized numerous engaging events and insightful workshops, making a meaningful impact on our student community.

This award is a testament to the collective effort, passion, and commitment of the entire team at Sairam IEEE EMB Society.

Acknowledgments

I extend my gratitude to our principal Dr. Porkumaran Karantharaj, HOD Dr. Priya E, and Dr. K. Sri Dhivya Krishnan for their unwavering support and mentorship throughout this journey.

2nd Place - IC Hack 2.0 National Hackathon

Our team, BYTE STORM , secured the second position at the IC Hack 2.0 national-level hackathon, earning a cash award of ₹10,000 .

Project Overview

Our project was developed for the Safety and Transportation track, where we applied Computer Vision techniques to address crucial challenges in the field.

This achievement was a testament to our team's collaboration, innovation, and dedication to creating impactful technology solutions.

Team BYTE STORM

• Saai Srivathsan
• Jayachandiran Kumar
• Kishore .K

2nd Runner-Up - Data Sprint Hackathon

My team and I secured the 2nd runner-up position in the "Data Sprint" hackathon, an intense 24-hour competition hosted by the Department of AI-DS at Sri Sairam Institute of Technology.

Project: Medical Chat AI

We developed a Medical Chat AI designed to assist users with health-related queries. The key features included: - Advanced disease detection through image processing algorithms. - Integrated text-to-speech support for accessibility.

This experience was a fantastic opportunity to push our limits, learn rapidly, and collaborate effectively under pressure.

Team

• Saai Srivathsan
• Kishore .K
• Jayachandiran Kumar

2nd Place - SDG Ideathon 3.0

Our team achieved 2nd place in the SDG Ideathon 3.0, a competition focused on creating solutions for the United Nations' Sustainable Development Goals.

Focus Area & Goal

Our project addressed Goal 6: Clean Water and Sanitation . We developed an innovative idea aimed at tackling challenges related to water quality and accessibility, leveraging technology for a sustainable impact.

This journey highlighted our team's collective passion for sustainability and our ability to work cohesively to develop meaningful solutions.

Team & Mentors

• Teammates : S. Nivedhitha, M.P. Sachin, Amathul Rifa
• Mentors : Dr. Priya E, Dr. Srinivasan Arunsankar Narayanan