Preprints
Meta-Learning for Cross-Task Generalization in Protein Mutation Property Prediction
Meta-learning framework enabling few-shot generalization across protein mutation prediction tasks.
Graph-Based Kirchhoff Modeling of Non-Ohmic Electron Transport in Self-Assembled Nanonecklace Networks
Graph-based modeling of electron transport in self-assembled nanonecklace networks.
Work history
Founding Engineer
- End-to-end ownership of ML, systems, and deployment for an AI-for-fintech product.
- Shipped through multiple 0→1 pivots across engineering, product, and go-to-market.
Machine Learning Engineer
- Multi-agent protein engineering pipelines integrating ESM, ProteinMPNN, AlphaFold, and RFdiffusion.
- Built a modular model zoo and LangChain interface connecting generative and predictive workflows for wet-lab scientists.
Graduate Researcher
- Multi-Peptide (JCIM 2024): Multimodal peptide-property predictor combining sequence and AlphaFold-generated structures.
- Multimodal Catalysis (Nature MI 2024): Generative LLM + contrastive transformer-GNN for adsorption energy without crystal-structure data.
- MOFGPT (JCIM 2025): RL-driven generative transformers for de novo MOF sequence design.
- Drug Discovery Agent (JCIM 2026): Agentic pipelines with LLMs, RAG, and property-prediction models.
- Meta-learning preprint for cross-task protein mutation prediction.
- Mentored junior researchers; founded DreamPods, an LLM+TTS science communication initiative.
MITACS Globalink Research Intern
- LSTM and Neural CDE surrogates for hydro-processing flow-rate optimization on real-time refinery data.
- Presented results at the Three-Minute Thesis symposium to 50+ faculty. Watch talk
Research Fellow
- ML surrogates for stiff chemical kinetics, reducing CFD simulation time by 90% for reactor design.
Machine Learning Intern
- Deployed image recognition on AWS SageMaker for automated appliance diagnostics from user photos.
Beyond the papers
Science communication, education tools, and applied research projects.
DreamPods
Founded an 11-person CMU team building LLM+TTS pipelines that turn scientific material into accessible audio.
GenAI Education Platform
PDF to lecture video pipeline using multimodal LLMs for automated STEM content creation.
Character-Consistent Diffusion
CMU Generative AI course Best Poster. Coherent visual storylines via latent alignment in diffusion models.
Refinery Process Optimization
MITACS @ UAlberta. LSTM/Neural CDE surrogates with 97% accuracy on real-time plant data.
Awards & service
ACS JCIM Peer Reviewer
Reviewing manuscripts on AI-driven molecular modeling and chemical informatics.
Oct 2025 to presentMITACS Globalink Scholar
International merit-based research scholarship at the University of Alberta.
2022Young Research Fellowship
Selected among top undergraduates at IIT Madras for early research.
Jul 2021Best Poster Award
Character-Consistent Diffusion, CMU Generative AI course.
Nov 2024