Building practical software and machine learning systems.
I build machine learning workflows, backend services, and data-driven applications that focus on usability, experimentation, and measurable impact.
Presenting Echoes of Longevity
Current focus
Applied machine learning and model evaluation.
Building GlassBoxML, a hand-implemented ML library with a full DevOps pipeline (Docker, CI/CD, Kubernetes).
Reproducible analytics workflows and experimentation.
About
Curious, analytical, and happiest close to the problem.
I like work that blends statistical thinking and mathematical intuition with dependable software. The goal is not just a notebook that works once, but a workflow someone else can understand, run, and trust.
I am a Computer Science student with a strong foundation in data science and practical software roles. My strongest work sits at the intersection of Python, machine learning, backend services (Java/Spring Boot), data manipulation, SQL, and product-minded communication.
I have developed end-to-end workflows ranging from automated credit scoring prototypes to extensive NLP research. Across all my projects, I focus heavily on the mechanics of the algorithms I use, ensuring the final predictive models are clear, deliberate, and trustworthy.
Skills
A practical toolkit for data and software delivery.
The stack is intentionally compact: tools that help move from raw data to validated insight, then into a clean prototype or service.
Machine Learning
Modeling, evaluation, feature engineering, and experiment tracking.
Data Engineering
Reliable pipelines for analysis-ready datasets and dashboards.
Software Engineering
Service-oriented APIs with practical Spring Boot exposure.
Cloud & Delivery
CI/CD pipelines and container orchestration for deployable project environments.
Projects
Selected work with a measurable learning goal.
Each project is designed to show practical judgment: problem framing, model choice, deployment awareness, and clear communication.
Echoes of Longevity: Healthy Ageing Narratives in Science, News, and Social Media
Large-scale NLP research analyzing how healthy ageing is discussed across PubMed, news media, Reddit, and YouTube. Applied transformer-based sentiment and emotion classification, interrupted time series modeling to detect lasting shifts in discourse around the COVID-19 pandemic, and Granger causality analysis to test lead-lag dynamics between platforms.
1st Place & Special Recognition Award — TDK Scientific Conference (2025); 2nd Place — EELISA Student Scientific Conference (2026).
Resource-Efficient Neural Networks: Quantization Sensitivity
Ongoing BME Project Laboratory research into how fixed-point quantization affects neural network training and robustness. Investigates post-training and quantization-aware training, gradient deadlock, gradient scaling, error accumulation, zero-order optimization, and a custom differentiable stair non-linearity for hardware-aware training, progressing from linear models to a quantized two-layer MLP.
Supervised BME Project Laboratory course — formal report with 7 controlled experiments; research in progress.
GlassBoxML
Classic ML algorithms (linear regression, logistic regression, decision tree) hand-implemented from scratch and exposed through an interactive FastAPI web demo with a Chart.js frontend.
Containerized with Docker, CI/CD via GitHub Actions and Jenkins, deployed to a local Kubernetes cluster (minikube).
Credit Risk Prediction Pipeline
End-to-end ML prototype for credit risk prediction, covering preprocessing, feature engineering, model training, and evaluation.
REST API endpoints for model inference, containerized with Docker.
Recognition
Recognized for the research behind Echoes of Longevity.
Award recognition presented at two academic conferences.
1st Place & Special Recognition — TDK BME Scientific Conference, 2025, for the Echoes of Longevity research.
2nd Place — EELISA Student Scientific Conference, 2026, for the Echoes of Longevity research.
Experience
Growing through coursework, labs, and independent builds.
A concise timeline of the experiences shaping my technical direction and collaboration habits.
Internship
Data Science Research Intern
HSDS-Lab
Designed and automated scalable Python ETL pipelines processing ~100k documents, and integrated LLM APIs for structured output generation and prompt engineering experiments.
Internship
Finance Data Science Intern
FinKell
Built a Python backend with REST APIs for an end-to-end credit risk prediction pipeline, containerized and deployed with Docker on AWS.
Internship
Java Software Developer Intern
FinanceSoft
Developed and maintained Java Spring Boot microservices in a production financial system, strengthening test coverage with JUnit and Mockito in an Agile/Scrum team.
Sep 2023 – June 2027
Computer Science Engineering BSc
BME
Coursework in algorithms, databases, machine learning, and software engineering fundamentals; top 10% of programme by GPA.
Contact
Open to internships, junior roles, and project collaboration.
The best fit is a team that values curiosity, careful analysis, and clean implementation.