At Mohur, I work on AI systems for investor-startup intelligence. I designed the investor-startup knowledge graph and benchmarked graph neural network architectures, improving link-prediction AUC from approximately 0.78 to 0.93.
I developed the Founder Readiness Index from the ground up, with sector-specific criteria and stage-adjusted scoring. I also used SHAP-based explainability to examine investor-startup matches and translate the findings into founder archetypes.
I designed and personally developed the initial prototype of Mohur's agentic copilot, including its frontend, chatbot, and hybrid recommendation system. I also lead the technical team through task allocation, product development, and deployment.
My research at Amrita Vishwa Vidyapeetham focuses on temporal graph learning, explainability, and agentic systems. Previously, at Kleaner Pyrotech, I developed reporting tools and predictive models, and analyzed purchasing and warehouse inventory processes that contributed to a 27% reduction in the cost-to-overhead ratio.
