Punnoose K Thomas

Slow Fashion Clothes Classifier -

A Multi-tier Deep Learning Model that focuses on classifying clothes into different categories based on their visual features on their images. Uses a comparison between different models to show how optimal the final suggested combination of models are.

Skills: Python, AI, Deep Learning, Data Analysis and Augmentation, Visualization, Model Comparison

Medical Chatbot -

A Nuxt.js frontend for the Mental Health Chatbot Application that interacts with the Flask backend. Uses transfer learning on DiabloGPT model using the mental health dataset from HuggingFace.

Skills: Python, Flask, NuxtJS, JavaScript, LLM, AI, Bash Scripting

Data Analysis -

Clean, analyze, visualize and infer from the given historical data how clients can be offered more alternate options of delivery based on urgency and price. Created Models to predict various values.

Skills: KNIME, Python, SQL

Graph Database -

Find an existing Ontology for movies. Populated the ontology with data from 3 different datasets on Kaggle. Wrote queries to fetch recommendations based on various paramteres and deployed them to a FastApi backend. Deployed HTML files as front-end using the same backend endpoints.

Skills: Python, RDF, FastAPI, HTML, CSS, JavaScript

Bus Tracker App -

Cross Platform Application for tracking college buses. Meant for parents to ensure students got to college safely.

Skills: C#/XML, Visual Studio, Android, iOS

Virtual Fence -

Image Processing AI program trained on YOLOv2 using ImageNet Dataset that monitors video feed from forest borders in Kerala to notify authorities when wild animals cross the border. Allows Forest officials to take action before nearby villages are attacked.

Skills: Python, Jupyter Notebook