Mudit Sinha

Accomplished AI Lead skilled in large language models, data science, software engineering. Expert in LLM frameworks, proven ability to lead AI initiatives end-to-end. Strategic thinker, analytical problem-solver. Innovative, detail-oriented, adaptable professional driven to better human experiences via AI.

• Business Analyst Specialization Wharton
• Machine Learning - Stanford

• Winner Calhacks (Community and Civic Engagement Award)
• Runner Up - BobHacks
• Winner ACM UCM Datathon

Alfred, August 2023-Current

• Created an intelligent personal assistant to assist me in my day to day life. It leverages a knowledge base of documents, including email templates, recipes, movies-to-watch, music to provide tailored assistance.
• Implemented an initial working version using the Mixtral 8x7B and RAG architectures, allowing the bot to effectively utilize my personal documents. Later, transitioned a smaller Llama 3 8B model as I wanted to experiment with fine-tuning.
• Project significantly improved my enjoyment of routine tasks, leading to overall greater producitivity.

Kyusei, April 2022

• Kyusei is a personalized tool for python developers and data science enthusiasts that take the popular static analysis tool Metabob API and makes it more user friendly.
• The project utilizes Bag-of-Words embeddings, NumPy, Matplotlib and Streamlit.
• The project aims to promote the idea that even young developers can use automated tools to debug and analyze errors in their code.

Continual Distributed Learning for Crisis Management, Priyanshu, A., Sinha, M. & Mehta, S., Accepted at the 3rd Workshop on Continual and Multimodal Learning for Internet of Things, IJCAI 2021, 04/2021, 05/2021

Alfred, August 2023-Current

• Developed intelligent AI assistant to streamline daily tasks by tapping into personal knowledge base - email templates, recipes, entertainment recommendations, and more.
• Initial implementation leveraged Mixtral 8x7B and RAG architectures for effective document utilization. Later migrated to Llama 3 8B model to explore fine-tuning capabilities.
• Project significantly improved my enjoyment of routine tasks, leading to overall greater productivity.

Kyusei, April 2022

• Kyusei is a personalized tool for python developers and data science enthusiasts that take the popular static analysis tool Metabob API and makes it more user friendly.
• The project utilizes Bag-of-Words embeddings, NumPy, Matplotlib and Streamlit. The project aims to promote the idea that even young developers can use automated tools to debug and analyze errors in their code.

Continual Distributed Learning for Crisis Management, Priyanshu, A., Sinha, M. & Mehta, S. Accepted - IJCAI 2021, 04/2021, 05/2021