Projects | Srivathsan Badrinarayanan

Featured Projects

Explore my research and development work in AI for science, generative models, and data science applications.

AI4Science

ACS JCIM

Multi-Peptide: Multimodality Leveraged Language-Graph Learning of Peptide Properties

Developed a multimodal deep learning model for peptide property prediction, combining sequence and structural data using contrastive learning. Achieved state-of-the-art results.

Paper GitHub

Nature Machine Intelligence

Multimodal Language and Graph Learning of Adsorption Configuration in Catalysis

Created a generative-predictive pipeline to predict adsorption energy for adsorbate-catalyst systems, combining LLMs and graph-based multimodal learning.

Paper GitHub

THALA - Thermodynamics Helper and Analysis Library for Applications

Released a Python library for thermodynamic analysis, supporting end-to-end project completion and version control.

GitHub

Generative AI / LLMs

Generative AI Podcasts

Developed a generative AI pipeline to create automated podcasts using LLMs and TTS models, enhancing public accessibility to knowledge.

Spotify

Automated Lecture Video Creation

Developed an automated ML pipeline to convert PDFs to lecture videos, reducing preparation time for educators and aiding efficient content creation.

Time Series & Machine Learning Modeling

Data-Driven Optimization of Refinery Process Operations

Built accurate multivariate deep learning models to optimize refinery operations, increasing profitability and operational efficiency.

Machine Learning Pipeline for Object Recognition

Designed a machine learning pipeline for object recognition deployed on AWS Sagemaker, reducing service wait times by 50%.