Hello World!
It's Atharv
Idevelopsoftware,applymachinelearning,andmakedata-drivendecisions.
AI Atharv
A bit about me.

My
Projects.

Driver Drowsiness Detection
Developed a real-time driver drowsiness detection system using computer vision and deep learning. The system monitors driver's facial features, particularly eye movements and head position, to detect signs of drowsiness. Implemented using OpenCV for face detection and custom CNN model for drowsiness classification, achieving 94% accuracy in real-world testing scenarios.

Modalytics
Developed an automated ML Model Evaluation Pipeline integrating AWS Lambda, S3, DynamoDB for model evaluation. Reduced model evaluation time by automated testing and performance evaluation done by deploying python model evaluation scripts in docker containers to AWS Lambda. Computed metrics such as MSE, MAE, R² Score and stored them in DynamoDB. Built an interactive streamlit dashboard to present the metrics over time in an intuitive manner for easy visualization

Wine Quality Prediction Pipeline
Developed an end-to-end ML pipeline for predicting wine quality using ElasticNet regression. The system features automated data processing, MLflow integration for experiment tracking, and a Flask web interface. Implemented comprehensive pipeline stages including data validation, transformation, and model evaluation, achieving high prediction accuracy with thorough performance metrics tracking.

AI Job Board
AI-driven Job Board leveraging data from US Bureau of Labor Statistics to provide actionable career insights. Integrated a salary prediction model with over 85% accuracy using Random Forest Regression and a user-based collaborative filtering recommendation system powered by cosine similarity, enhancing suggestions by 30%. Implemented an interactive dashboard with the streamlit and plotly libraries, to visualize key trends such as salary projections, high-demand skills, growing and declining industry sectors, and much more, enabling users to make data-informed career decisions.

Autonomous Driving Model
A Convolutional Neural Network (CNN) for autonomous driving, inspired by Nvidia's end-to-end learning for self-driving cars research paper, tested using the Udacity Driving Simulator. Achieved a 90% success rate in lane keeping and applying data augmentation techniques such as Gaussian Blur to improve training results. Attained 98% accuracy on the model after training on 50 epochs using the Google Cloud Platform.

Clash Royale Analytics Dashboard
Built an interactive analytics dashboard using python and the panel library to efficiently process and analyze over 13 GB of player data from the popular strategy game Clash Royale. Leveraged algorithms like K-Means Clustering to uncover insights on metrics such as elixir usage, card usage and other gameplay trends across 20+ arenas, assisting strategic decision making for players.
Work
Experience.
Machine Learning Research Assistant
Rutgers University
- Implemented a Multiple Convolution Network (MulCNN) for classification of Heat Shock Proteins into distinct protein families, achieving a 94% accuracy.
- Processed and transformed 12,000+ data points from FASTA files into pandas dataframes, optimizing data preprocessing for seamless model training.
- Developed a web-based tool integrating the model to provide real-time classification for users.
Software Developer Intern
Independence Education
- Created a comprehensive data analytics dashboard that links to the Canvas student software for an Ed-Tech startup, leveraging TypeScript, React.js, Chart.js, PostgreSQL, and Node.js to provide actionable insights
- Engineered a powerful chatbot prototype utilizing the GPT-3.5-turbo model and LangChain, integrating a Retrieval- Augmented Generation (RAG) architecture for enhanced accuracy and contextual relevance.
- Employed advanced data visualization techniques to present complex datasets in an intuitive manner, significantly enhancing user engagement and enabling data-driven decision-making for improved learning outcomes for both, learners and educators
- Employed advanced data visualization techniques to present complex datasets in an intuitive manner, significantly enhancing user engagement and enabling data-driven decision-making for improved learning outcomes for both, learners and educators
Software Engineering Intern
VIIE
- Developed microservices using Node.js and deployed them using Docker containers.Spearheaded the development of the VIIE Web Application utilizing cutting-edge technologies including React.js, Next.js, and Tailwind CSS ensuring scalability, performance, and maintainability of the application and delivering a rich and engaging user experience.
- Successfully managed a diverse, 15-member cross-functional team, including Frontend, Backend, Database, and UI/UX specialists. Fostered a collaborative environment, facilitating effective communication and cooperation between developers and stakeholders.
- Successfully managed a diverse, 15-member cross-functional team, including Frontend, Backend, Database, and UI/UX specialists. Fostered a collaborative environment, facilitating effective communication and cooperation between developers and stakeholders.
- Actively engaged with stakeholders to gather requirements, provide progress updates, and incorporate feedback, ensuring the final product met or exceeded expectations.
Machine Learning Intern
FeynnLabs
- Piloted the successful implementation of diverse Machine Learning Models, including advanced techniques like LSTM, Multiple Linear Regression, and Support Vector Regression effectively to solve complex optimization challenges, resulting in enhanced operational efficiency and approximately 15% growth in revenue.
- Conducted data exploration and generated comprehensive reportsthat provided actionable insights, empowering stakeholders to make informed decisions and advance strategic initiatives
- Conducted an in-depth analysis of the electric vehicle market, identifying three major growth opportunities that prompted a strategic shift resulting in 25% more investment towards innovative EV initiatives over six months.
- Collaborated effectively with cross-functional teams, including data scientists, engineers, and business analysts, to ensure the successful deployment of machine learning models. Facilitated knowledge sharing and fostered a collaborative environment to drive innovation and continuous improvement.