Deep Learning Image Classification
A custom CIFAR-10 architecture and controlled Fashion-MNIST ablation study implemented in PyTorch.
Best CIFAR-10 test accuracy87.81%
Project 01
Graduate portfolio
Aaron Gulzar
MSc Data Science, Queen Mary University of London
Machine LearningData ScienceSoftware Engineering
I build and evaluate practical machine-learning systems, with work spanning computer vision, information retrieval and software development. My approach combines scientific rigour with clear, maintainable implementation.
01 / Featured Projects
Technical work, presented with evidence.
Academic experiments and software projects with transparent status, methods and limitations. Exact metrics are shown only where preserved results support them.
Information Retrieval
A search-ranking project exploring BM25 scoring, text preprocessing and interpretable retrieval results.
Core ranking methodBM25
Heart-Disease Model Transportability Audit
Investigating how heart-disease classifiers behave across clinical datasets, with an emphasis on careful validation, reliable probabilities and reproducible analysis.
Research focusModel reliability
Stud Safari
A Roblox zoo-management project focused on gameplay systems, Lua scripting and interactive design.
Engineering focusGame systems
Project 04
local habitat = Zoo:CreateHabitat()
habitat:AddAnimal(animal)
economy:UpdateRevenue()
player:UnlockExpansion()
TradingView Indicator Suite
A developing suite of chart indicators organised around reusable signal logic and clear visual output.
Current phaseDevelopment
Project 05
indicator("Signal Suite", overlay=true)
trend = ta.ema(close, length)
signal = ta.crossover(close, trend)
plot(trend)
Portfolio Website
A responsive, accessible static portfolio built without a framework or runtime dependencies.
Frontend dependencies0
Project 06
<main> <section id="projects"> Semantic, responsive content </section> </main>
02 / About
Scientific foundations, applied through engineering.
I am studying for an MSc in Data Science at Queen Mary University of London after completing a BSc (Hons) in Biochemistry at Sheffield Hallam University.
My scientific background shaped how I approach machine learning and artificial intelligence: define the question carefully, control the experiment, inspect the evidence and communicate limitations. I am developing that foundation through software-engineering practice, with an interest in building practical systems that are reliable, understandable and useful.
03 / Skills
Repository-backed technical skills.
Grouped by how they are used across university work and current repositories.
Programming
- Python
- SQL
- Lua
- HTML, CSS & JavaScript
Machine Learning
- PyTorch
- scikit-learn
- Convolutional neural networks
- Model evaluation
Data Science
- pandas
- NumPy
- Matplotlib
- Information retrieval
Software Engineering
- Git & GitHub
- Modular design
- Responsive interfaces
- Technical documentation
Tools
- Jupyter Notebook
- Google Colab
- VS Code
- Roblox Studio
04 / Education
Academic background.
Postgraduate data science training built on an undergraduate scientific education.
Queen Mary University of London
MSc Data Science
Postgraduate study focused on data science, machine learning and related computational methods.
Sheffield Hallam University
BSc (Hons) Biochemistry
A scientific foundation in biological systems, experimental reasoning and analytical interpretation.
05 / Experience
Work and community experience.
Presented at the level currently supported by the available record.
Tesco
Retail Experience
Customer-facing retail work in a team environment, requiring reliability, clear communication and attention to day-to-day operational responsibilities.
Community
Volunteering
Volunteering experience contributing time and support to community activity. Specific responsibilities and dates will be added once verified.
06 / Contact
Open to graduate machine-learning, data and software roles.
For opportunities or project discussions, email is the most direct way to get in touch.