A selection of engineering automation and data pipeline projects delivered for manufacturing, R&D, and operations teams.
This standalone executable interfaces directly with the SolidWorks API to programmatically automate thermal flow simulation setup. Design engineers who previously spent hours manually configuring simulation parameters for each assembly can now trigger the entire process with a single click. The tool scans assemblies, queries a material database, applies the correct flow simulation properties to each component, and handles edge cases like fasteners and missing data — all automatically.
A serverless data pipeline triggered by massive XML file exports from the Bluestar product configurator. Python scripts running on Azure Functions parse the raw XML, fully reconstruct complex 150% Bills of Materials, and load the structured output into SQL databases for downstream reporting. The system includes real-time monitoring through Microsoft Teams notifications for errors, validation issues, and data integrity checks — and feeds directly into Power BI dashboards.
A strict Medallion architecture implementation where data flows sequentially through Bronze (raw JSON ingestion), Silver (cleansed and validated Delta Parquet tables), and Gold (business-level star schema aggregations) layers. The entire pipeline is orchestrated using Python Notebooks within Microsoft Fabric, with the Gold layer feeding Power BI reports via Direct Lake mode for near-instant query performance.
An internal web application and REST API built for a global lighting manufacturer. The system serves as a centralized hub for engineering data, providing programmatic access to retrieve, convert, and manage critical technical specifications and component data. It eliminates manual lookups across disconnected spreadsheets and legacy systems, giving engineering teams a single source of truth accessible from any internal tool or script.
An automated deal-aggregation platform for Canadian parents, built on FastAPI. Scheduled scrapers collect product promotions across multiple retailers, then an AI layer writes bilingual (English/French) product descriptions, filters out items that aren't relevant to the target audience, and curates a daily selection of the best deals. Everything is published automatically to a bilingual WordPress storefront — with no manual copywriting or editorial work.
An AI assistant we built and use internally at Drakarian — proof of the same approach we design for clients. Claudette connects a large language model directly to our day-to-day tools (calendar, task lists, email, and a knowledge base), so it can take instructions in plain language and act on them: planning the day, drafting emails, logging work, and recalling project context across sessions. It runs as a self-hosted service that exposes its tools through the Model Context Protocol (MCP) — the same architecture we build for clients who want AI wired into their own systems.
A full-stack Django web application with user authentication and custom forms for dynamically calculating CO2 emissions based on user-input data.