"Your skills are not just tools, but potential. To truly unlock your potential, you must use your skills with purpose and intention."
Unity-based Robot Training with ML Agents and Reinforcement Learning
I developed a project using Unity, ML Agents, and reinforcement learning to train a virtual robot.
By combining simulation and machine learning, the robot learned to navigate towards a target location autonomously. I employed advanced techniques like reward shaping and policy optimization to optimize the training process.
This project showcased my proficiency in Unity and demonstrated my ability to solve complex problems in robotics and AI.
​
Technologies Used :- Unity, ML-Agents, Reinforcement Learning, Imitation Learning
Face Generation using Generative Adversarial Models
I developed an innovative Anime Face Generating Model that utilized a state-of-the-art Generative Model Architecture and advanced machine learning techniques, resulting in high-quality results and optimized outputs.
Additionally, I enhanced the model's performance by integrating a Deep Convolutional Generative Adversarial Network with Wasserstein Loss and conducted a comparative analysis of the results generated by both models.
​
Technologies Used :- TensorFlow, Keras, pandas, numpy, matplotlib, PIL
Fine Tuning Stable Diffusion 1.5
Developed a CLI application using Vertex AI, GCP to generate AI images using Stable Diffusion, for text-to-image generation.
Enabled customizable image creation through a Custom Diffuser Fine Tuned style (Kerala Mural Painting) using Accelerate.
Leveraged advanced techniques like Self-Attention Guidance and ControlNets to produce high-quality output images
​
​
VerboFad Web Application
Developed and deployed a cutting-edge web application utilizing the state-of-the-art Transformer (T5) mode. The application empowers users to summarize, paraphrase, and analyze similarity scores between two texts, thereby enhancing their productivity and efficiency.
Created an intuitive and user-friendly interface for the web application using industry-standard technologies such as HTML, CSS, Bootstrap, and JavaScript. Implemented the web application controllers using Flask Web Application routers and HTML hyperlinks, enabling seamless navigation between web pages and enhancing the user experience.
​
Technologies Used :- nltk, pandas, numpy, Flask, HTML, CSS, Bootstrap, Javascript
Smart Bin Monitoring System
Developed and executed an efficient Intelligent System for bin monitoring, utilizing sensor technology and internet connectivity to provide comprehensive insights. Incorporated various activators and microprocessors, including the ESP32 Microprocessor, to ensure optimal system performance with low latency.
Streamlined communication between the system and the database with minimal latency by leveraging the secure MQTT Protocol and a managed cloud service, AWS.
​
Technologies used :- C, C++, Arduino, ESP32
​
​
Movie Recommendation System
The movie recommendation system is an AI-powered tool that helps users discover movies based on their viewing history and preferences.
It uses a collaborative filtering approach to make movie recommendations that best match the user's interests. The system analyzes the user's historical data, such as watched movies, movie ratings, and search history, to suggest personalized movies.
​
Depression detection from social networks using ML and DL techniques
Developed a natural language processing-based intelligent machine learning system for depression detection in individuals using Twitter data, and optimized its performance through comparative analysis of machine learning and deep learning models
Leveraged natural language processing techniques such as lemmatization, stop word removal, and TF-IDF vectorization to attain an 80% precision rate.
​
Technologies used :- nltk, pandas, numpy, scipy, seaborn, sklearn, itertools
