Shivesh Prakash

About Shivesh Prakash

I am a MFE candidate at UC Berkeley (Haas) and an incoming Quant Summer Associate at J.P.…

Experience

1. Undergraduate Researcher

   Middleware Systems Research Group Msrg

   Sep 2023 — Present · Toronto, ON, CA

   A Deep Learning Framework for Step-Wise Synthesis Planning via Data-Driven Exploration - Developed MHNpath, a machine learning-driven retrosynthetic tool leveraging modern Hopfield networks to enhance computer-aided synthesis planning (CASP)- Introduced a user-tunable scoring system to prioritize greener, cost-effective pathways based on reaction cost, temperature, and toxicity- Validated MHNpath through experimental synthesis and case studies with pharmaceutical compounds, demonstrating its potential to discover novel, sustainable synthetic routes.(Under review at Nature Machine Intelligence)2. Equivariant Graph Neural Network for Rapid Transition State Structure Predictions in Chemistry - Designed GNNStruc, a graph neural network (GNN) for predicting chemical reaction transition state structures with SE(3) symmetry, enabling accurate molecular representations- Achieved competitive accuracy while significantly reducing computational costs compared to traditional quantum mechanical methods- Demonstrated its scalability on datasets with reactions, paving the way for high-throughput drug design and catalysis discovery.(To be submitted to the Journal of Chemical Information and Modeling)

Education

• University of Toronto

  Honors Bachelor of Science, Computer Science

• University of California, Berkeley

  Master's degree, Financial Engineering

• Narayana Junior College - India

  Class 12 CBSE, Science

• The Orchid School

  Class 7

  2010 — 2017

• Narayana Olympiad School

  Class 10

• Scottish High International School

  Class 1

  2008 — 2010