Computational Materials Physics & AI

I am completing my PhD in Computational Materials Physics under Prof. Dario Alfè at University College London (thesis submitted), specializing in first-principles calculations, atomistic modelling, machine learning, and agentic AI systems for autonomous materials discovery. My research develops ab initio methodologies for high-pressure binary alloy phase equilibria and melting curves, integrating DFT, molecular dynamics, and machine-learned interatomic potentials with thermodynamic free-energy calculations. I model material defects, electrides, and photocatalysts via Gibbs free-energy reaction pathways. Furthermore, I compute electronic, structural, mechanical, magnetic, optical, and quantum-transport phenomena via DFT and NEGF formalisms.

18+
Publications
209+
Citations
10
h-Index (Scholar)
13+
Invited Peer Reviews

Recent Research Output Highlights

Core Scientific & Computational Areas

Research Thrusts Thrusts 1–3 of 5
01

First-Principles Electronic Structure

Kohn-Sham DFT solving electronic ground states, anisotropic auxetic mechanics, and dynamical phonon stability in novel 2D lattices.

VASP Phonopy HSE06
02

High-Pressure Phase Equilibria & Melting

Ab-initio molecular dynamics (AIMD) and thermodynamic integration predicting binary alloy melting curves under extreme planetary core conditions.

AIMD LAMMPS Free Energy
03

Materials Informatics & Graph Neural Networks

Rigorous replication audits of crystal graph models (CGCNN, MEGNet, SchNet) uncovering target-feature leakage and establishing valid property benchmarks.

PyG CGCNN Audit Protocol
04

Autonomous Agentic DFT Workflows

Neurosymbolic architectures deploying local 4B LLM multi-agent crews for literature parameter extraction, deterministic validation, and Slurm self-correction.

Local LLMs Slurm SIESTA
05

Solar Photocatalysis & Materials Discovery

High-throughput screening and 2D materials discovery, resolving strain-engineered band alignments and thermodynamic reaction pathways for solar photocatalysis.

PDC Band Align HER/OER

Selected Publication Highlights

Ab initio High-Pressure Phase Diagrams of Al-Mg Alloys
THRUST 01 · HIGH-PRESSURE PHASE EQUILIBRIA Figure 1 of 7

Ab Initio High-Pressure Phase Diagrams of Al–Mg Alloys in the Low Solute Limit

Fully ab initio DFT and thermodynamic free-energy integration mapping the dilute-limit Al–Mg binary phase diagram up to 150 GPa. Uncovers a pressure-driven partitioning reversal above ~60 GPa where Al transitions from liquid-favouring to solid-favouring during the Mg hcp-to-bcc transformation, inverting the coexistence field topology.

METHODOLOGIES & CODES
DFT / VASP Thermodynamic Integration AIMD Phase Coexistence Extreme Conditions
arXiv:2608.02225 (2026) →

Academic Background & Experience

Research & Academic Appointments
UCL
PhD Candidate · Doctoral Researcher
University College London · London, UK
Dec 2022 – Present Thesis Submitted 2026
  • Formulated ab initio frameworks to compute high-pressure phase diagrams of binary alloys.
  • Designed and automated HPC workflows for melting curve and coexistence determination.
  • Parameterized and validated empirical and machine-learning interatomic potentials (MLIPs).
  • Investigated pressure-induced solid–solid polymorphism and mechanical stability from first principles.
  • Quantified premelting phenomena, defect energetics, and thermomechanical properties of alloys.
  • Modeled water-splitting electrocatalysts, electride systems, and defects with experimental teams.
DFT AIMD LAMMPS Python & Fortran
KU
Research Associate
Khalifa University · Abu Dhabi, UAE
Jan 2022 – Nov 2022 · 11 mos
  • Executed high-throughput DFT screening to discover and predict functional 2D materials.
  • Investigated facet-dependent surface energetics and electronic structure for photocatalytic applications.
  • Mentored postgraduate researchers in computational project design and workflow execution.
2D Materials Photocatalysis Water Splitting VASP
Education & Credentials
UCL
Doctor of Philosophy (PhD), Condensed Matter Physics
University College London · London, UK
2022 – 2026 Thesis Submitted
DOCTORAL DISSERTATION

Ab initio Methods for High-Pressure Phase Diagram Calculations within the Low-Solute Limit

PRIMARY ADVISOR
Prof. Dario Alfè · Thomas Young Centre · University College London
TU
Master of Science (MSc) in Physics
Tribhuvan University · Kathmandu, Nepal
Graduated 2020 First Class with Distinction
MASTER'S THESIS

Ab initio Calculations of Structural, Electronic and Quantum Transport Properties of 2D Gold Monolayer

PRIMARY ADVISOR
Prof. Rajendra Adhikari · Kathmandu University
Applied Data Science & Machine Learning Specializations
Coursera · DeepLearning.AI · AWS · IBM · CrewAI
2019 – Present Online Specializations

Industry-certified online specializations and technical curricula covering deep learning architectures, statistical machine learning, cloud computing & MLOps (AWS), enterprise data science (IBM), and autonomous multi-agent workflows (CrewAI & DeepLearning.AI).

DeepLearning.AI Coursera AWS IBM CrewAI PyTorch & scikit-learn MLOps

Technical Expertise & Computational Stacks

Electronic Structure & Quantum Transport

Ab initio calculations with VASP, SIESTA, tranSIESTA, DFT+U & NEGF methods

VASP SIESTA tranSIESTA Quantum ESPRESSO DFT + U NEGF Methods HSE06 Hybrid DFT Phonopy Band Topology

Molecular Dynamics & Statistical Mechanics

Classical & ab initio MD (LAMMPS, AIMD), free-energy methods & coexistence

Classical MD (LAMMPS) AIMD (VASP) Thermodynamic Integration Coexistence Techniques Free-Energy Methods Phase Boundaries & Melting Thermodynamic Ensembles Embedded Atom Model (EAM)

ML Potentials & Atomistic Modelling

MACE, DeepMD-Kit, MAML validation & Pymatgen, ASE, OVITO, VESTA analysis

MACE (Equivariant) DeepMD-Kit MAML Potential Training & Validation Pymatgen ASE OVITO VESTA Active Learning Loops

Machine Learning & Deep Learning

End-to-end predictive models with PyTorch, TensorFlow, scikit-learn & Keras

PyTorch TensorFlow scikit-learn K Keras Graph Neural Networks (PyG) PINNs (Physics-Informed) Surrogate Models XGBoost & Trees End-to-End Pipelines

Generative AI & Autonomous LLM Agents

Agentic multi-LLM systems, RAG pipelines, LoRA fine-tuning, LangChain & APIs

Multi-Agent LLM Systems LangChain LangGraph CrewAI RAG Pipelines LoRA Fine-Tuning LLM APIs Local LLMs (4B) MCP Protocol Deterministic Code Gates

Scientific Computing, HPC & Cloud MLOps

Linux clusters (SLURM, PBS), Docker, AWS (SageMaker, Bedrock), Terraform & ETL

Python 𝔽 FORTRAN Bash SQL & NoSQL Vector Databases Linux Clusters (SLURM, PBS) Docker AWS (SageMaker, Bedrock) Terraform Kafka & Airflow (ETL) Pandas & Spark Plotly & Streamlit Git / GitHub LaTeX Jupyter & Colab Antigravity IDE & VS Code

Ongoing Projects & Research Frontiers

THRUST 01 · AUTONOMOUS SCIENTIFIC AI Active Platform · Extending arXiv:2608.29665
Project 1 of 3

Autonomous 2D Photocatalyst Discovery Engine (2D-PhotoCatLab)

Autonomous multi-agent discovery & high-throughput screening of direct-gap 2D monolayers for solar water splitting.

Autonomous 2D Photocatalyst Discovery Engine (2D-PhotoCatLab)
Click to enlarge (Full 4K)
RESEARCH OVERVIEW & SCIENTIFIC NOVELTY

Extending our local open-weights LLM workflow (Sharma, arXiv:2608.29665) beyond reproducibility audits into an end-to-end autonomous discovery laboratory. Guided by the strict neurosymbolic rule that "agents propose, deterministic code disposes", lightweight local SLMs operate via Ollama with logit-level GBNF grammar constraints, orchestrating generative composition, MLIP pre-screening, autonomous self-healing DFT, and rigorous physical stability gates with zero API metering and zero IP leakage.

METHODOLOGIES & COMPUTATIONAL CODES
5-STAGE NEUROSYMBOLIC DISCOVERY PIPELINE

Publications

01 2026
arXiv · arXiv:2608.29665

Evaluating a 4B open-weights local LLM for agentic DFT workflows: a literature reproducibility audit

S. B. Sharma*

02 2026
arXiv · arXiv:2608.02225

Ab initio High-Pressure Phase Diagrams of Al–Mg Alloys in the Low Solute Concentration Limit

S. B. Sharma*, S. Mehta, D. Alfè

03 2026
Appl. Surf. Sci. · Vol. 744, 167411

First-principles electronic structure and photocatalytic water splitting under uniaxial strain in pentagonal PdTe₂ monolayer

N. Kumar, S. B. Sharma, D. Alfè, R. Pandey, N. Tit

04 2026
J. Chem. Inf. Model. · Under Review

Comment on "Hybrid Graph-Machine Learning Framework for Accurate and Interpretable Band Gap Prediction"

S. B. Sharma*

05 2025
J. Chem. Phys. · Vol. 162, 184502

Ab initio phase diagrams of binary alloys in the low solute concentration limit

S. B. Sharma*, S. Mehta, D. Alfè

06 2024
Mater. Horiz. · Vol. 11, 3613–3622

Signature of low dimensional quasi-F center in zirconium rich oxide

D. Thapa, S. B. Sharma, D. Alfè, D. Kilin, S. Kilina

07 2024
J. Phys.: Conf. Ser. · Vol. 2726, 012014

New 2D penta-SiPN: A wide and indirect bandgap semiconductor

S. B. Sharma*, I. Qattan, S. KC, S. Abedrabbo

08 2023
Physica E · Vol. 146, 115517

Structural deformation and mechanical response of CrS₂, CrSe₂ and Janus CrSSe

S. B. Sharma*, R. Paudel, R. Adhikari, G. C. Kaphle, D. Paudyal

09 2023
Comput. Mater. Sci. · Vol. 218, 111952

First-principles DFT study of structural, electronic and optical properties of Cu-doped TiO₂ (112) surface for enhanced visible-light photocatalysis

S. B. Sharma, I. Qattan, M. Abi Jaoude, S. Abedrabbo

10 2022
ACS Appl. Electron. Mater. · Vol. 4, 2561–2569

Penta-SiCN: A highly auxetic monolayer

S. B. Sharma*, I. Qattan, M. Jaishi, D. Paudyal

11 2022
Appl. Surf. Sci. · Vol. 593, 153239

Enhanced optoelectronic and elastic responses in fluorinated penta-BN₂

S. B. Sharma*, S. KC, D. Paudyal

12 2022
Appl. Surf. Sci. · Vol. 599, 153997

Enhanced mechanical, electronic, magnetic and optical properties of 2D CrX₂

S. B. Sharma*, I. Qattan

13 2022
ACS Omega · Vol. 7, 36235–36243

Large Negative Poisson's Ratio and Anisotropic Mechanics in Penta-B₂C

S. B. Sharma*, I. Qattan, S. KC, A. Alsaad

14 2022
Carbon Trends · Vol. 7, 100162

Strain dependent electronic and optical responses of penta-BCN monolayer

S. B. Sharma*, R. Bhatta, R. Adhikari, D. Paudyal

15 2022
Nanomaterials · Vol. 12, 4068

First-Principles Prediction of New 2D p-SiPN: A Wide Bandgap Semiconductor

S. B. Sharma*, I. Qattan, S. KC, S. Abedrabbo

16 2022
Asia Pac. Phys. Conf. · pp. 73–79

Highly Tunable Negative Poisson's Ratio in Two-Dimensional Penta Structures

S. B. Sharma*, I. Qattan, S. KC

17 2021
Eur. Phys. J. B · Vol. 94, 128

Structural, electronic, magnetic and mechanical properties of Cr₂X₃ (X=S,Se,Te)

S. B. Sharma*, R. Bhatta, K. R. Sigdel, R. Adhikari, G. C. Kaphle

18 2021
Chem. Phys. Lett. · 2021

Optoelectronic and elastic response of fluorinated hexagonal boron nitride

S. B. Sharma*, R. Bhatta, R. Adhikari

19 2021
Carbon Trends · 100162

Strain driven negative Poisson's ratio and extraordinary light absorption in 2D penta-CN₂

S. B. Sharma*, R. Bhatta, R. Adhikari, D. Paudyal

20 2021
J. Nepal Phys. Soc. · Vol. 7, No. 1

Strain induced electronic and optical properties of 2D silicene

S. B. Sharma*, R. Bhatta, R. Adhikari, G. C. Kaphle

View Complete Publication Record on Google Scholar →

Contact

GET IN TOUCH

Open to research collaborations, academic discussions, and opportunities in computational physics & scientific AI.

NAME
Shambhu Bhandari Sharma
PhD Researcher · UCL Physics & Astronomy
EMAIL
Direct academic & research inquiries
UCL PROFILE
Official UCL IRIS Profile & Directory