Harvey Bellini.

Education

University of Exeter

MSci Computer Science (2020-2024)

  • 8+ years of Python experience in programming and software development
  • Java-based OOP with real-world applications
  • Web Development: Created a Covid-19 web tracker using HTML, CSS, JavaScript, and PHP
  • Database Design: Implemented SQL databases for company applications
  • Network and Security: Secured websites from XSS and SQL injection using cryptography
  • AI and Nature-Inspired Computing: Solved TSP using ACO and particle swarm optimization
  • Mobile Computing: Built an Android app that suggests outfits based on real-time weather
  • AI Pathfinding: Applied graph traversal and path search algorithms in AI projects
  • Machine Learning: Utilized ML techniques to produce statistical analysis reports
  • Software Engineering Project Leader: Led a team of 6 using Scrum, achieving the highest mark in the cohort

Master's Dissertation - Visualising the Search Paths of Evolutionary Algorithms

  • Developed novel visualisation techniques (VTs) for Multi-Objective Particle Swarm Optimization (MOPSO).
  • Implemented MOPSO in Python and applied it to five diverse test problems: ZDT3, DTLZ1, DTLZ2, WFG3, and DF2.
  • Proposed five VTs: Animation, PCA, Wind Map, Coloured Wind Map, and Colour-Mapping, evaluating their effectiveness using custom evaluation criteria.
  • Key contributions include the Wind Map VT, which provides both individual particle and overall algorithmic trajectories in a single figure, enhancing decision-making.
  • Used Gestalt principles and a taxonomy of usage to evaluate the cognitive accessibility of each VT.
  • Future work focuses on expanding evaluation criteria with real-world problems and enhancing accessibility for decision-makers (DMs).

Bachelor's Dissertation - Optimising Reinforcement Learning for Asteroids: A Comparative Analysis of RL Algorithms with Hyperparameter Tuning

  • Conducted a comparative analysis of reinforcement learning algorithms (A2C and PPO) for playing the Atari game Asteroids
  • Evaluated the impact of CNN vs MLP neural network architectures on agent performance
  • Investigated the effects of frame stacking and number of parallel environments on training efficiency and final scores
  • Performed hyperparameter tuning, focusing on learning rate and discount factor (gamma)
  • Implemented and trained models using Python and the Stable Baselines 3 library
  • Achieved a maximum score of 901.7, improving upon random play baseline
  • Analyzed training data to identify rapid initial learning followed by performance plateau
  • Discussed limitations of the approach and potential areas for future work, including use of more computational resources and alternative algorithms

Graveney School

A-Level: Computer Science, Maths, Physics (2018-2020)

Work Experience

The Co-operative Group

Customer Team Member - Exeter, 2023

  • Practiced effective communication with both customers and colleagues on a daily basis
  • Efficiently integrated with the existing team and made a quick impact

Exeter Entrepreneurs

Workshop Leader - University of Exeter, 2022

  • Led a 10-week Python course teaching a range of university students
  • Learned key presentation and communication skills teaching to a variety of abilities

Deliveroo

Rider - Exeter, London, 2020-2022

  • Communicated with customers and restaurants professionally, resolving issues efficiently
  • Optimized routes and selected orders to ensure timely deliveries

CeX

Retail Sales Assistant - London, 2019-2020

  • Recommended consumer electronics to enhance customers' quality of life
  • Answered product questions with up-to-date knowledge of sales and store promotions
  • Trained new associates on customer transactions and balancing the cash drawer

Personal Projects

F-Vision

Developed bespoke 3D software for Manchester City Football Club over 10 weeks.

CoLab Mental Health System

Led a team of six to build custom software for an Exeter-based charity supporting individuals with complex issues.

Took charge of client communication to further my skills in effective communication with non-technical stakeholders.

Developed several key features including database (MySQL), and backend (Python).

Wetherspoons Best Drink

Created a tool to find the best price per unit drink at Wetherspoons.

Used the undocumented API and data processing pipeline to calculate the "best" drink.

Spotify to Tidal

Developed a command-line tool to import Spotify playlists into Tidal.

Programmed using both the Spotify and Tidal API to sync music.

Co-op Shifts Calendar Sync

Synced Co-op shifts with Google Calendar integration.

Automated the process, meaning all upcoming and previous Co-op shifts were synced with Google Calendar automatically.

Used Docker for managing the application.

Stable-baselines3

Contributed to SB3, an open-source machine learning library used by thousands.

Ensured high-level documentation was accessible to all users.

Django Boiler Template

Created a Django template for easy deployment of web projects like this one.

Allows users to spin up instances within seconds.

Contact Information

harveybellini@gmail.com