Kajol Agrawal

Dedicated System Analyst with a solid foundation in computer science and extensive experience in machine learning and deep learning. Holds a Bachelor's degree in Computer Science and a Diploma in Data Science. Proven ability to develop predictive models and implement robust applications, enhancing system reliability and minimizing downtime through innovative solutions.

Music Generation using Deep Learning

• Built a character-level LSTM based Recurrent Neural Network (RNN) for the sequence modeling and generation of musical data.
• Implemented sequence-to-sequence training and utilized embedding layers, softmax classifier to predict note sequences from ABC notation and MIDI inputs
• Optimized model performance through batch processing and epoch-based training, focusing on robust data preprocessing techniques.

Taxi Demand Prediction

• Built a predictive model for taxi demand, forecasting the number of pickups in specific regions within a 10-minute interval to provide actionable insights to drivers
• Managed and preprocessed a massive dataset utilizing Dask for memory-efficient block-by-block loading of files, engineered new features from raw 19 fields, focused on outlier removal, and did exploratory data analysis on them
• Developed and compared Time series regression models, optimizing performance using Random Grid search, and evaluating with MAPE and MSE

Youtube Chatbot using Langchain

• Built an end-to-end Retrieval-Augmented Generation (RAG) pipeline using Langchain to build a Youtube Chatbot, focusing on leveraging Large Language Models (LLMs) to generate answers.
• Integrated Youtube API for transcript ingestion, implemented an Embedding model for vectorization and designed a custom Retriever to execute semantic search against Vector Store.