I build Python systems for problems that start with raw data, manual workflows, or an idea that needs to become a dependable product.
My work sits between backend engineering and applied AI. That can mean building APIs and data pipelines, designing retrieval systems, or putting the infrastructure around a model so that it can actually be used and maintained.
I care about the parts that are easy to skip in a demo: clear data models, reliable services, validation, documentation, and deployment. I have backgrounds in computer science and applied statistics, which helps me look at both the technical system and the data behind it.
What I work with
Backend and data
Python · FastAPI · Django · Flask · PostgreSQL · MySQL · SQLAlchemy · Redis · REST APIs · GraphQL · ETL pipelines · Data validation
Applied AI
LLMs · Fine-tuning · RAG · pgvector · Ollama · Llama · LoRA · PyTorch · scikit-learn · spaCy · BERTopic
Cloud and delivery
Docker · AWS · Lambda · API Gateway · S3 · CloudFormation · GitHub Actions · CI/CD
A few things I’ve built
FastAPI RAG Engine
A retrieval-augmented generation backend built with FastAPI, PostgreSQL with pgvector, and Ollama. It supports asynchronous processing, semantic chunking, streamed responses, authentication, and multi-tenant data access.
The project is intentionally focused on the service around the model, not just the chat response. It is a practical look at the API, database, retrieval, and deployment concerns behind a RAG application.
Mindful Eating Companion
A personalized digital intervention that combines behavioural data, a mobile application, backend services, and AI-assisted guidance. I worked on the FastAPI backend, PostgreSQL data layer, personality-based personalization, and LLM integration.
It was a useful project for thinking through how a model fits into a real product, including data flow, validation, API design, and the needs of the mobile client.
Public Tenders Dashboard
A visual analytics project built around public tender data from Nova Scotia. The work involved preparing and exploring the data, then using clustering and BERTopic to make patterns in a large collection of records easier to investigate.
The goal was simple: turn raw public data into something people could explore, question, and use.
More than a stack
I started with competitive programming, where the job is to understand constraints before writing a solution. That habit still carries into my work. I like getting close to the actual problem, asking what can go wrong, and building systems that are easier for the next person to understand and work with.
Get in touch
I am based in Halifax, Nova Scotia, and open to conversations about backend engineering, data systems, and applied AI work.