Dishan Bhandari

Download CV
Bio
An avid learner driven by intellectual curiosity and continuous self-improvement. My background in Data Science equips me with a strong foundation in data analytics, statistics, machine learning, and deep learning, complemented by the ability to apply data-driven approaches to operational challenges, process improvement, and decision-making.


2025 - 2026
MSc in Data Science from the University of Greenwich with Distinction (76.67%). My master's dissertation, titled "Explainable Early Dropout-Risk Prediction Study for Students" investigated a temporally constrained machine learning methodology for early student dropout-risk prediction using real-world data from the Open University Learning Analytics Dataset (OULAD). The study prioritised explainable AI (XAI) to improve the interpretability of model predictions and employed temporal constraints to prevent data leakage by ensuring that predictions relied only on information available at the time of prediction.

Alongside my core modules, I undertook elective modules in Blockchain for FinTech Applications and Technologies for Anti-Money Laundering and Financial Crime.
2024 - 2025
Selected for the DataCamp Donates Fellowship Program 2024/25, ranking among the top 1% of learners on the platform. Completed Tracks on: ML Engineer, Associate Data Engineer and Data Analyst
2025  
Officially registered as a Licensed Computer Engineer under Nepal Engineering Council.
2019 - 2024
Bachelor of Computer Engineering from Pokhara University with CGPA of 3.44/4.0. My final year project was "Fake Currency Detection using Deep Learning", which involved photographing counterfeit banknotes under The Central Bank of Nepal (NRB)'s controlled access arrangements. The collected images were subsequently used to construct a custom dataset for developing a fake currency detection system which utilized Convolutional Neural Network.
Other Projects
All these projects are listed with their respective github links but you can check my github profile GitHub for other minor projects as well.

UoG Courseworks repo includes all my coursework specifications with reports and notebooks across various course modules that I submitted as part of my Masters assessments. These should be used morally and ethically as a reference rather than plagiarism.
In Car Price Prediction, I developed and deployed a containerized machine learning REST API using FastAPI to serve regression predictions, with authentication and authorization for endpoint access. Integrated Redis for response caching and Prometheus with Grafana for runtime metrics monitoring and visualization. Used Docker to containerize the application and its supporting services, and deployed the API on Render with interactive API documentation.
Since I loved reading books, why not build a recommender that will help me find my next read. Semantic Book Recommender is an intelligent, user-friendly book recommendation system powered by a pretrained text classification LLM model which takes in account about the user's preference for various categories and the emotional relevance they are seeking in a book. The recommendations were displayed through an interactive web application dashboard using gradio.
I built PDF Chat Assistant while doing my Bachelor's when I needed fast and convenient summaries for different lecture PDFs when preparing for board exams. It utilizes Agentic RAG that allows users to chat naturally, ask questions about PDFs (research papers or academic files) while providing the functionality to generate summaries related to that PDF, all in a single interface. I used Groq Llama 3 for LLM responses and built it with: LangGraph, LangChain and LangSmith with Streamlit for frontend deployment.
While taking DataCamp's Transformers course, I came across the revolutionary "Attention Is All You Need" paper from Ashish Veshwani et al. So I decided to try it out by building a simple sequence to sequence translation model, coding the Transformer architecture from scratch.
ICE Detention Analysis was actually a screening task sent to my friend. The tasks were to clean and preprocess the dataset, compute total population per detention facility and finally, identify and visualize the top 10 largest facilities by population. Since it was in R, I decided to give it a shot myself as well.
Statement of Accomplishments

Beyond the data
  • I absolutely love football and I am a die-hard Manchester United fan. So, you can imagine how mentally resilient I am!
  • I am into video games as well, especially Valorant and Football Manager.
  • I have my own nerd side as well. I enjoy watching movies, series, anime, reading novels and listening to music.