Harshal Soman Data Scientist

About · Projects · Resume
Avatar

I'm a graduate student pursuing my Masters in Computer Science at University of Illinois at Chicago. Before I came to the USA I obtained my Bachelors in Electronics and Telecommunication Engineering from University of Mumbai. Passionate about Data Science, Machine Learning and Artificial Intelligence. I love working with great people, getting creative with data, using state-of-the-art technologies and building awesome products. Interested in working together or having a chat? Feel free to contact me.

Analyzing

In a wide range of subject areas, I have analyzed structured and unstructured data to extract actionable business insights. I love to craft stunning and clever visualizations that illustrate surprising results.

Developing

I'm strongly convinced that machine learning models should not go to waste in Jupyter Notebooks. Using my software engineering skills, I've built and deployed AI services which create real business value.

Communicating

I enjoy public speaking, writing professional articles, sharing my knowledge and discussing diverse topics. Thanks to my training and experience in science communication, I'm able to present complex results to a non-technical audience.

Featured Projects

wrting

RAG based Apple Bot

The Apple Financial Insights Chatbot is an efficent tool designed to provide real-time answers to financial questions related to Apple Inc. Leveraging the power of advanced language models, including Google Flan-T5 and Llama via llama-cpp, the chatbot utilizes SEC filings to deliver precise and insightful responses. By implementing a Retrieval-Augmented Generation (RAG) system, the chatbot efficiently combines dense retrieval with state-of-the-art language models to enhance the relevance and accuracy of its answers.

Check it out on Github

wrting

Railway Crack Detection Bot

The Railway Crack Detection Bot uses advanced machine learning techniques to identify and report cracks in railway tracks in real-time, ensuring enhanced safety and reliability in railway transportation. The model is trained on a manually obtained dataset, providing high accuracy in detecting potential hazards.

Research paper Check it out on Github

wrting

Sentiment Analysis of Covid-19 Tweets

This project involved analyzing public sentiment on Twitter regarding the Omicron variant and the potential second lockdown in India. Data was manually collected using a Tweepy crawler, and sentiment trends were visualized to understand public opinion.

Manuscript Check it out on Github

wrting

Wind Power Prediction using Explainable AI

Developed a predictive model to forecast wind power generation using historical data and machine learning algorithms. The project emphasized model explainability to understand factors influencing wind power generation, enhancing the efficiency and reliability of wind energy systems.

Check it out on Github

wrting

Covid-19 Detection using cough audios

This project utilizes a Convolutional Neural Network (CNN) to detect COVID-19 infections from audio signals like cough and breath sounds. By preprocessing audio data and training the CNN model, the project achieves an accuracy of 87.55% on the test set.

Check it out on Github

wrting

Another Signal Clone

Created a secure messaging application using the Signal protocol for end-to-end encryption, ensuring private communication. The project encompasses both server and client components, developed in Python, providing a reliable platform for confidential messaging.

Check it out on Github

mplcyberpunk

GitHub Projects

I'm hosting a number of smaller projects on GitHub which includes a cool website for Age Recognition,a Telegram bot and a Money manager app. Feel free to have a look around.

Check it out