SCIENTIFIC AI
COMPUTATIONAL BIOPHYSICS
Understanding proteins.
Building intelligent models.
I explore the intersection of protein biophysics, machine learning and quantum computing to develop physics-grounded representations of molecular systems.
Ipswich, UK / PhD researcher · University of Suffolk
THE RESEARCH THREAD
I am a PhD researcher in computational protein biophysics, machine learning and quantum-computing modelling, nearing completion of my doctoral thesis. My work connects protein dynamics, molecular simulation and scientific computing.
Before my PhD, I built full-stack applications for cheminformatics, bioinformatics and production ML. My aim is to bring AI and quantum computing together in hybrid drug-discovery pipelines that support better decisions in molecular design, simulation and optimisation.
My path from engineering to research01 / RESEARCH
Three connected directions in protein modelling, scientific AI and high-performance computing.
A research framework representing proteins through residue-level physical features, interaction structure and sampled state distributions.
Exploring how quantum circuit sampling and thermodynamic reasoning can offer interpretable views of protein energy landscapes and residue interaction networks.
A high-performance Python library for protein solvation analysis, using analytical and numerical solvent-accessible surface area (SASA) algorithms with parallel optimisation enabled by JAX.
Studying how solvation-aware molecular features can improve the interpretation of protein conformational ensembles and molecular dynamics trajectories.
The related manuscript, JaxProtSASA, is listed as in press.
GNN- and transformer-based manifold learning, with residue-level and atom-level graph representations for analysing protein conformational flexibility and ensemble behaviour.
Working with molecular dynamics-derived data structures using Python, PyTorch, PyTorch Geometric, NumPy, SciPy and MDAnalysis.
Presented research talks at Evotec on protein conformational ensembles and graph-based ML representations.
Ongoing research is described at an abstract level. Further technical details are available for discussion.
02 / EXPERIENCE
From scientific prototypes to deployable applications, and from molecular simulations to learning representations.
Evotec Pharma · Industry funded
Research in AI-driven molecular modelling, computational protein biophysics and quantum computing. Developing physics-grounded ML representations of protein conformational ensembles, residue interactions, solvent-mediated effects and molecular state transitions using Markov state models.
Research directions include physics-informed neural networks, scalable molecular dynamics post-processing, ensemble interpretation, druggable-state selection and simulation-informed drug discovery.
e-Zest Solutions Limited
Translated research prototypes and domain-specific requirements into maintainable services, interfaces and deployment workflows for cheminformatics, bioinformatics, retrosynthesis, sequence analysis and ML-assisted applications.
Helped move scientific ML and molecular informatics workflows from prototypes towards user-facing, reproducible and deployable software.
Developed components for retrosynthesis route prediction, reaction-tree visualisation, molecular editing and prediction serving.
Integrated RDKit, AiZynthFinder-style workflows, TensorFlow, Flask, Angular, MarvinJS and Docker in a multi-container scientific application.
Developed a web-based sequence analysis platform with distributed job execution for ML-driven clustering, alignment and sequence processing.
Built backend services and job orchestration using FastAPI, Celery, RabbitMQ, Redis, MySQL and VueJS.
Built and deployed prototypes across computer vision, NLP and recommendation workflows, developing practical experience in model deployment, APIs, containers and production ML design.
03 / PUBLICATIONS
Preprints, work in press and book chapters across molecular computing and applied AI.
arXiv · 2608.05491
In press
In press
In A. Heifetz (Ed.), High Performance Computing for Drug Discovery and Biomedicine. Methods in Molecular Biology, vol. 2716. Humana, New York, NY.
In P. Thealla, V. Nadda, S. Dadwal, L. Oztosun & G. Cantafio (Eds.), Corporate Cybersecurity in the Aviation, Tourism, and Hospitality Sector (pp. 258–272). IGI Global Scientific Publishing.
04 / BLOGS
A space for short articles on protein modelling, scientific machine learning and quantum computing.
THE NOTEBOOK
In the meantime, explore the projects and publications behind my research.
Explore the research05 / SKILLS
Connecting physical reasoning, machine learning and dependable scientific software.
Graph neural networks, transformers, representation learning (Data2Vec), physics-informed neural networks, information-theoretic analysis, group theory and manifolds.
Protein conformational ensembles, molecular dynamics analysis, sequence analysis and retrosynthesis.
Quantum state representation and quantum machine learning, connected to molecular modelling and ensemble analysis.
Full-stack scientific applications, distributed job processing, cloud-native applications and deployment workflows.
06 / EDUCATION
FEB 2023 — PRESENT
Digital Futures Institute
University of Suffolk
Ipswich, UK
THESIS
Quantum Computing and AI methods for Protein Ensemble modelling
Industry-funded research fellowship by Evotec Pharma.
AUG 2015 — MAY 2019
Computer Engineering
Savitribai Phule Pune University
Pune, India
Projects in reinforcement learning for a neon car racing AI agent, NLP feature extraction, and applied machine learning for health and societal applications.
Production Machine Learning Systems Google Cloud · 2023
End-to-End Machine Learning with TensorFlow on GCP Google Cloud · 2022
BEYOND THE DESK
FIRST PRIZE · 2026
University of Suffolk
Protein Encoding on Quantum Circuit for Ensemble Analysis
Three-year PhD scholarship awarded by Evotec Pharma, 2023.
Presented the Protein Encoding on Quantum Circuit poster at the MGMS conference.
Sponsored participation to build and benchmark Fragment Molecular Orbital (FMO) binding-affinity calculations using GAMESS (US) on AMD Instinct MI300A accelerators.
Presented at Quantum.Tech, London.
07 / GALLERY
A space for photographs, posters and moments from research life.
Gallery images will appear here when added.
08 / GET IN TOUCH
For conversations about protein modelling, scientific AI, quantum computing or research opportunities.
p.patil@uos.ac.ukBASED IN
Ipswich, United Kingdom