Abin Shakya

ML Researcher | Data Scientist | Algorithmic Trader | Mathematics Enthusiast

Home Projects Blogs Download CV

Profile

I am a PhD student in the Department of Computer Science and Engineering at Louisiana State University. My research interests lie in designing graph-based neural networks to model complex relationships and interactions inherent in various physical systems. Additionally, I am passionate about creating robust and reliable AI models that not only deliver accurate predictions but also quantify the uncertainty associated with these predictions. I enjoy building trading algorithms for fun. It is my way of trying to teach my computer to earn its keep around here--so far, it hasn't asked for a raise!

Skills and Research Interests

Programming/Frameworks Python Django R C++ CUDA Matlab
ML/Data Science frameworks Pytorch Tensorflow Keras Pandas PySpark Scikit Learn
Research Interests Graph Learning Representation Learning Large Language Models Generative AI XAI Time Series Analysis
Research Domain of Interest Molecular Dynamics Finance / Quantitative Research Physics Mathematics Language Vision
Other Skills/Interest Algorithmic Trading Software Engineering Pattern Recognition DevOps Linux Latex

Professional Experience

Graduate Research Assistant - Louisiana State University

August 2021 - Present

  • Developed a weakly supervised Graph Neural Network for scalable 3D phase segmentation in molecular dynamics simulations, enabling generalization from small systems to large-scale atomic structures, and authored the paper "Weakly Supervised Graph Neural Networks for Scalable 3D Phase Segmentation in Molecular Dynamics Simulations." [Accepted for Oral Presentation at KDIR 2025]
  • Working on geometric generative modeling to create precise atomic configurations, maintaining structural integrity and compliance with essential physical constraints.
  • Building Graph Neural Network-accelerated Molecular Dynamics to explore atom distribution in Earth's core.
  • Designed an active learning workflow for molecular dynamics, utilizing iterative data sampling and model training cycles to continuously enhance model accuracy.
  • Developed deep learning-based potential energy models for molecular dynamics, facilitating simulations of Earth's interior, and authored the paper "Insights into core-mantle differentiation from bulk Earth melt simulations." DOI: 10.1038/s41598-024-69873-8
  • Conducted binning analysis to identify bimodal distribution and verify core-mantle segregation using strongly connected components. Also created a 3D convex hull to model the Earth's core boundary, employing various methods for validation.
  • Developed a transformer-based generative AI model that composes original piano music, crafting expressive and dynamic melodies for an immersive listening experience.
  • Implemented a state-of-the-art Bayesian Neural Network architecture to enhance prediction uncertainty in Neural Networks, contributing to more reliable and robust predictions.
  • Developed indexing algorithms that empower efficient queries on compressed data, resulting in optimized data retrieval processes. This innovation directly contributed to the increased speed and effectiveness of data analysis.

Graduate Teaching Assistant - Louisiana State University

January 2023 - Present

  • Teach lab sessions and assist students with programming assignments for Computer Science I – Java (CSC1351).
  • Grade and conduct office hours for the course : Applied Deep Learning (CSC4343).
  • Teach and Grade the course : Introduction to Python (CS2700).
  • Grade and conduct office hours for the course : Data Science and Analytics (CSC2730).
  • Grade and conduct office hours for the course : Interactive Computer Graphics (CSC4356)

Senior Software Engineer - Sparrow Pay Private Limited

October 2019 - August 2021

  • Build a middle-ware layer called Service Layer that integrates APIs from multiple vendors (both SOAP and REST API) and expose a consistent REST API for our clients.
  • Managed continuous integration and deployment (CI/CD) processes, overseeing the DevOps lifecycle including staging deployments and configurations for client-side testing.
  • Carried out a complete upgrade and transition of the codebase from Django 1.8 to Django 2.2, integrating modern web development techniques to boost system stability.
  • Developed a custom task scheduler using the Celery framework to automate essential backend operations, such as generating financial reports, reconciling transactions, and sending periodic transaction summaries via email to clients.
  • Enhanced the performance of a flight search engine through strategic implementation of advanced caching techniques and parallel programming.
  • Implemented secure file uploads to AWS S3 using presigned URLs to improve data security and optimizing access efficiency in cloud storage environments.
  • Collaborated closely with the finance and accounts departments to analyze and rectify discrepancies in financial documentation, enhancing accuracy and compliance.
  • Offered mentorship and technical guidance to junior software engineers, directing project strategies and promoting professional growth within the team.

Software Engineer - Sparrow Pay Private Limited

October 2017 - October 2019

  • Developed client facing and admin user interface in Angular JS.
  • Build Rest APIs using Django Rest Framework to perform CRUD operations(Flight Booking , Hotel Booking, Electricity Bill Payment etc).
  • Integration of Web SDK to facilitate online payments.
  • Collaborated closely with the Operations team to ensure the provision of essential data required for their workflows.
  • Address and resolve technical issues reported by Customer Support.
  • Document APIs in Swagger.

Education

PhD in Computer Science

Louisiana State University - August 2021 - Present (4.062/4.0)

Bachelors of Computer Engineering

Tribhuvan University - November 2013 - September 2017 (3.602/4.0)

Publications

  • Shakya, A., Karki, B.B. (2025). Weakly Supervised Graph Neural Networks for Scalable 3D Phase Segmentation in Molecular Dynamics Simulations. Accepted at KDIR 2025.

  • Shakya, A., Ghosh, D.B., Jackson, C., Morra, G., Karki, B.B. (2024). Insights into core-mantle differentiation from bulk Earth melt simulations. Scientific Reports. DOI: 10.1038/s41598-024-69873-8.

  • Shakya, A., Pokharel, A., Bhattarai, A., Sitikhu, P., Shakya, S. (2018). Real-Time Stock Prediction Using Neural Network. 8th International Conference, Confluence 2018— IEEE. ISBN: 978-1-5386-1720-5.