Advay Kadam

WildHacks II

• Won the Best Environmental Hack prize in a 48-hour Hackathon focused on creating technology that addresses the environment, sustainability, and wildlife endangerment.
• Trained a convolutional neural network (80% accuracy) using Tensorflow and created a web-application using Flask, HTML, and CSS to develop a software that classifies whether an item is recyclable or trash.
• Utilized OpenCV so that the user can capture an image in real time and feed it into the CNN and used the Pyttsx3 Python library to return the model's results to the user through voice commands.

PioneerHacks III 

• Won the Best Artificial Intelligence Hack prize in a 24-hour prompt-free Hackathon
• Trained a neural network (70% accuracy) on 10,000+ images using Tensorflow to classify whether an image of the skin is cancerous or not.
• Developed a web-page using Flask to allow the user to input an image of the skin and run the classification model.

COVID-19 Variant Prediction (https://colab.research.google.com/drive/1erwLDiBXxJlkUhCGX8hBZt2PylqL7Yzi?usp=sharing)

• Used global genomic sequences data from the NCBI database to train a multinomial logistic regression model (90% accuracy) that could classify the region of origin of future COVID-19 strains given their genome.
• Feature engineered genomic data using Numpy by traversing strands of DNA sequences and tracking proper nucelobase pairs and mutations.
• Worked in a team of 4 under the guidance of a UC Berkeley graduate and presented the project to over 100 students and AI experts.
• Technologies used: Python, Pandas, Numpy, SciPy, Scikit-learn, Bio-python, tqdm

Skin Disease Classification Tool (https://github.com/Advay-K/skin_disease_detector)

• Trained over 20,000 images of data to develop a convolutional neural network (76% accuracy) using tensorflow.keras to classify a skin lesion into 1 of 10 categories (ex. Eczema, Atopic Dermatitis, Acne, Warts, etc).
• Utilized OpenCV to access a camera and process the user's image in realtime; Made a web interface to return the model's results and provide essential information on the disease that was classified.
• Technologies used: Python, Tensorflow, Scikit-learn, Pandas, Numpy, OpenCV, HTML, CSS

Python for Data Science and Machine Learning Bootcamp