cv
Basics
| Name | Shreyas Vinaya Sathyanarayana |
| Label | AI Research Scientist |
| shreyas.college@gmail.com | |
| Url | https://shreyasvinaya.github.io/folio/ |
| Summary | AI Research Scientist specializing in AI for Chemistry, Drug Discovery, and Molecular Machine Learning. Led the development of DeepRetro, achieving state-of-the-art results in retrosynthesis. |
Work
- 2025.08 - Present
AI Research Scientist
MStack AI
Designing novel Chemistry-first AI models to tackle Molecule Synthesis and Chemical supply chains. Led the design and development of chemical foundational models for chemical reaction understanding.
- Chemical Foundation Models
- Molecule Synthesis
- 2024.06 - Present
Visiting Researcher Intern
Yale University
Conducting research on molecule generation models using geometric deep learning with Prof. Smita Krishnaswamy.
- Geometric Deep Learning
- Molecular Design
- 2023.12 - 2025.03
Research Intern
Harvard Medical School
Investigating the neural mechanisms of language processing using AI models under Prof Gabriel Kreiman.
- Neural Language Processing
- Machine Learning for Neuroscience
- 2023.06 - 2025.08
Research Fellow (DeepRetro Lead)
Deep Forest Sciences
Led the design and development of DeepRetro: a Chemist's AI Co-Pilot for Collaborative Retrosynthesis.
- Achieved 96.3% success rate on USPTO-190 benchmark
- Winner of $100,000 Standard Industries Chemical Innovation Challenge
- Migrated DeepChem models from TensorFlow to PyTorch, increasing user base by 92%
Volunteer
- 2024.05 - Present
Mentor
DeepChem - Google Summer of Code
Mentored contributors on polymer generative pipelines and model migration from TensorFlow to PyTorch.
- Mentored Debasish Mohanty on polymer generative pipelines
- Mentored Bibhusundar Mohapatra on Chemception model migration
- 2022.08 - 2023.05
ML Lead
Google Developer Student Clubs, BITS Goa
Guided 25+ ML enthusiasts, conducted weekly training sessions, organized hackathons.
- Led weekly ML training sessions
- Organized hackathons and mentored in ML model building
Education
Awards
- 2024.12.01
Standard Industries Chemical Innovation Challenge Winner
Standard Industries
Won $100,000 for DeepRetro - AI-powered retrosynthesis framework
- 2024.04.01
Publications
-
2026 ChemBERTa-3: An Open Source Training Framework for Chemical Foundation Models
Royal Society of Chemistry's Digital Discovery
Open source framework for training chemical foundation models.
-
2026 DeepRetro: Retrosynthetic Pathway Discovery using Iterative LLM Reasoning
Nature Scientific Reports
Novel hybrid AI framework for retrosynthesis achieving 96.3% success rate on USPTO-190 benchmark.
-
2024 Open Source Molecular Processing Pipeline for Generating Molecules
NeurIPS 2024 ML4PS Workshop
Open-source pipeline for molecular generation.
Skills
| Machine Learning | |
| Deep Learning | |
| Graph Neural Networks | |
| Large Language Models | |
| Generative AI |
| Drug Discovery | |
| Retrosynthesis | |
| Molecular Generation | |
| Property Prediction | |
| Computational Chemistry |
| Tools & Languages | |
| Python | |
| PyTorch | |
| TensorFlow | |
| Git | |
| AWS | |
| Azure |
Languages
| English | |
| Fluent |
Interests
| AI for Science | |
| Drug Discovery | |
| Materials Science | |
| Computational Biology |
| Open Source | |
| DeepChem | |
| Scientific Computing |
Projects
- 2023.06 - 2025.08
DeepRetro
AI Co-Pilot for Collaborative Retrosynthesis using iterative LLM reasoning. Achieved state-of-the-art 96.3% success rate.
- Nature Scientific Reports Publication
- $100,000 Challenge Winner
- 2023.01 - Present
DeepChem
Core contributor to the open-source drug discovery toolkit. Migrated models from TensorFlow to PyTorch.
- 92% increase in user base
- GSoC Mentor 2024-2025