CV

Doctor of Philosophy (PhD)

Cornell University

Systems Engineering

• Thesis: Thermochemistry-aware Machine Learning for Sustainable Molecular Design: Representation Learning, Radical Design, and PFAS Remediation
• Advisor: Dr. Fengqi You

Master of Science (MS)

Cornell University

Chemical Engineering

• Thesis: Probing Ion Effects In Nanoconfined Aqueous Electrolytes: A Molecular Dynamics Study Using Neural Network Potentials
• Advisor: Dr. Shuwen Yue

Bachelor of Technology (BTech)

Indian Institute of Technology (IIT) Roorkee

Polymer Science; Minor: Applied Mathematics

• Thesis: A Kinetic Model for the Direct Thermal Liquefaction of Pine Wood
• Advisor: Dr. Shushil Kumar
• Class Rank: 2

PhD Researcher

Department of Systems Engineering, Cornell University

• Designed a BDE-conditioned transformer for generative molecular design, achieving distributional control over radical bond strengths with 81-94% validity and 84-92% novelty across BDE targets.
• Achieved up to 90% R² improvement in molecular property prediction by distilling SchNet and DimeNet++ into 2× smaller student GNNs across QM9, ESOL, and FreeSolv datasets.

MS Researcher

Department of Chemical Engineering, Cornell University

• Trained DFT-accurate neural network potentials via active learning on HPC clusters for nanoconfined electrolytes, reaching force RMSEs ≤ 0.07 eV/Å and enabling fast MD simulations.
• Identified ion-specific interfacial behavior in confined electrolytes: K⁺ showed 3× stronger adsorption and 2× faster diffusion than Na⁺, trends absent in classical force fields.

Undergraduate Researcher

Department of Chemical Engineering, IIT Roorkee

• Developed a kinetic model for pine wood liquefaction, finding distillate formation nearly 2× faster than heavy residue, with secondary reactions negligible.
• Demonstrated that adding 20 wt% water more than doubled primary reaction rates, with guaiacol:water = 8:1 giving the highest overall conversion.

Research Intern

Department of CSE, Jadavpur University

• Boosted vehicular smoke detection accuracy by up to 12% mAP across 3 public datasets using a λ-attention transformer head on a YOLOv5 backbone.
• Increased training coverage by 5× with a dual-level synthetic smoke generation pipeline to overcome limited real-world data.

Research Intern

Department of Chemical Engineering, Indian Institute of Science

• Achieved 97% R² for nanopore formation probability and 95% R² for formation time prediction by training a two-stage ML framework on 20,840 unique graphene nanopore structures.
• Quantified the effect of 18 structural features on pore formation kinetics using SHAP-based feature importance, providing interpretable physical insights.

Best Bachelor's Thesis Award

IIT Roorkee

Received for developing a kinetic model for the direct thermal liquefaction of pine wood, later published in Biomass Conversion and Biorefinery.

Editor's Pick - Journal of Chemical Physics

Journal of Chemical Physics

Awarded for “Tailoring nanoporous graphene via machine learning” - selected as Editor’s Pick in the Journal of Chemical Physics.

Graduate Teaching Assistant - Investigative Biology Lab (BIOG 1500)

Cornell University

Taught 35 undergraduates per semester covering experimental design, hypothesis testing, and statistical analysis through hands-on labs.

Graduate Teaching Assistant - Fundamentals of Physics II Lab

Cornell University

Guided students through lab experiments ranging from electricity and magnetism to optics.