Hi,
I'm Mohammed Anas A R
AI/ML Enthusiast
About
I'am Mohammed Anas
Passionate Computer Science Engineering student with hands-on experience in full-stack development, AI/ML, and real-world software solutions. Skilled in Python, Java, JavaScript, Dart, and frameworks like MERN, Flutter, and TensorFlow, I ’ve built and optimized scalable applications and AI-powered systems through hackathons, academic projects, and internships. With a strong foundation in problem-solving and a growing focus on generative AI, prompt engineering, and cloud technologies, I ’m driven to build impactful solutions that address real-world challenges. I thrive in collaborative environments and am constantly exploring new technologies to grow as an innovative and reliable developer.
Works
She Space
"She Space" is a Booking Management System developed for the Trivandrum Corporation as part of the Smart City Initiative. The system serves as a public-facing and administrative portal that enables users to book facilities across four distinct regions,each with its own custom configuration. I developed a scalable backend using Node.js, Express.js, and MongoDB,and later migrated the system to MySQL for optimized performance. The app was initially deployed on AWS and is publicly accessible through a custom domain.Additionally, I integrated a secure government payment gateway to ensure a smooth and compliant transaction experience. The system is continuously maintained for uptime and stability, and the project received public recognition, including a feature in major newspapers and an acknowledgment from the Honourable Minister of Kerala. Github
Groundwater Prediction System
This AI-powered platform is designed to monitor and forecast daily groundwater levels in Coimbatore and other coastal regions. It features an administrative dashboard for government officials and a public interface for real-time data visualization. I trained a CNN-LSTM deep learning model using over 15 years of historical data to accurately predict groundwater trends. Real-time data from official government APIs, including rainfall, temperature, and groundwater measurements, are integrated to ensure up-to-date forecasts. A critical component is the automated alert system, which sends real-time warnings to officials if levels drop below sustainable thresholds. This project highlights the fusion of machine learning, real-time analytics, and civic technology to promote smarter water resource management. Github
Pump Demand Forecasting
This system is an AI-driven forecasting platform developed to assist MSME pump manufacturers in predicting monthly water pump demand across regions in India. I used 20 years of historical data—including rainfall, groundwater, and population metrics—to train a robust CNN-LSTM model that predicts upcoming production needs. The system supports both administrators and public users through a comprehensive web interface that includes visual trend analysis, demand forecasts, and alerts. Advanced feature engineering and preprocessing methods were employed to merge multiple datasets from government sources. The platform is currently under review by the MSME department for integration into official workflows, aiming to reduce production inefficiencies and resource misallocations. Github
Real-Time Sports Win Predictor
This system provides real-time sports analytics to predict win probabilities during live games using logistic regression. It was developed as part of my internship work and leverages live match data streamed through CSV files. The core algorithm factors in momentum shifts by analyzing recent performance trends using NumPy, Pandas, and the deque data structure. Python’s Watchdog library was used to track live updates and trigger re-evaluation of the win probabilities with every new event. The tool was tested on real-world match datasets and showed strong performance in forecasting game outcomes as the match progressed. Github
Tennis Ball Tracking System
I developed a real-time tennis ball detection and tracking system using TrackNet V3 architecture. This project focused on computer vision techniques to analyze player performance and ball motion in sports environments. I collected and annotated a custom dataset using RoboFlow and trained a specialized model to detect and track tennis ball movement under varying court conditions. Speed tracking was achieved by calculating displacement across video frames, making it a comprehensive analytics tool for tennis matches. The system was rigorously tested in live scenarios and validated for use in sports analytics and player coaching applications.
Hawk-Eye Badminton System
Inspired by the Hawk-Eye technology used in international sports, I helped develop a computer vision system for automatic court line detection and shot validation in badminton matches. This involved using Hough Transforms and a customized TrackNet model to detect shuttlecock paths and determine whether a shot landed inside or outside the boundaries. We validated the system under different lighting and court conditions to improve its reliability and accuracy in real-time decisions. This project contributed significantly to building integrity tools for sports analytics and officiating.
Flipkart Review Analyzer
This ML-powered system performs sentiment analysis and fake review detection on Flipkart product reviews. I built a classification model to detect whether reviews are positive, negative, or neutral, and a separate model to identify spam or fake content. Techniques included NLP, frequency analysis, and linguistic pattern recognition. The system was trained and validated on real-world datasets to ensure high accuracy and relevance to e-commerce platforms. Github
WhatsApp Generative AI Bot
An AI-powered chatbot integrated with WhatsApp that uses the LLaMA 3.2 language model running locally via Ollama. It answers user queries in real-time using LangChain for prompt chaining and Flask as a backend API. Built with Node.js and Baileys for WhatsApp automation, and Python for LLM processing. This is a full-stack Generative AI project demonstrating local inference, natural language understanding, and real-time messaging. Github
Sight Mate
Sight Mate is an assistive device for the visually impaired that uses a YOLO-based model to recognize Indian currency. I led a team to refine the system inherited from a senior batch, improving both software and hardware aspects. The object detection model was trained on a custom dataset for improved accuracy, and the compact hardware was optimized for speed, energy-efficiency, and usability. The project is officially patented and undergoing final testing for widespread deployment.
AI Resume and Cover Letter Builder
This project is a generative AI system that helps users automatically create resumes and cover letters based on their input skills, experience, and job interests. I developed a custom LLM backend that processes user input and dynamically generates tailored documents from structured templates and mapped skill-role datasets. The platform includes a frontend interface for easy data entry and instant document export, aiming to reduce the friction in job application processes.
E-Learning Management Website
I developed a scalable e-learning platform as part of a cloud internship, using the MERN stack and deploying it on AWS using EC2, S3, and RDS. The platform includes features like course creation, enrollment tracking, user dashboards, and JWT-based authentication. RESTful APIs and CI/CD pipelines were implemented to ensure a seamless development workflow. A/B testing and debugging were conducted to improve performance and usability. Github
Tourism Website with AI Planner
Built during the Smart India Hackathon Prelims 2024, this tourism platform features an AI-powered trip planner and real-time language translator. Tourists can input preferences and receive customized itineraries. The backend was built using FastAPI and Node.js with MongoDB support, while the frontend leveraged JavaScript and Bootstrap for responsive design. Github
NFA-Based Puzzle Game
I designed a Java-based educational puzzle game using the principles of Non-deterministic Finite Automata (NFA). The GUI, built with Java Swing, allows players to interactively test word inputs against a state machine, making it both fun and academically relevant. Github
AI Chatbot
I built a Python-based chatbot using the OpenGPT API, aimed at answering user questions across educational and casual topics. It includes a clean web panel and is trained on a custom dataset for enhanced contextual understanding. This project showcases the practical power of large language models in creating conversational interfaces. Github