Operations ResearchToolkit
The engineering and operations-research work behind the scenes — turning messy operational data into models that schedule, forecast and decide, packaged into reusable, explainable tools.
Overview
This is the analytical thread that runs through my engineering work: taking real operational data — messy, incomplete, coming from systems that were never meant to talk to each other — and turning it into models that actually inform a decision. Forecasting demand, scheduling resources, balancing flows, and quantifying trade-offs that were previously argued from gut feel. The goal is never a clever model for its own sake; it’s a tool someone can run, understand, and trust.
What I built
- Resource-flow analysis — modelling the flow of resources through an industrial process (including work on the Hellisheiði geothermal plant during my time at ON Power) to find where capacity and losses actually sit.
- Forecasting & optimization models — linear and mixed-integer programs, simulation, and statistical forecasting applied to scheduling and planning problems.
- Data plumbing — the unglamorous half: document-processing scripts, API-key integrations, and SQL/Power BI pipelines that get scattered data into one place clean enough to model.
- Explainable packaging — wrapping the analysis so the output is a decision aid a non-specialist can read, not a black box.
Why it’s hard
Operations research lives or dies on the data and the framing. The math is the easy part; the hard part is choosing the right model for a fuzzy real-world problem, getting trustworthy inputs out of imperfect systems, and presenting the result so people will actually act on it. It demands engineering judgment as much as optimization theory — knowing when a simpler model that ships beats an elegant one that doesn’t.
What’s next
Consolidating the recurring patterns — forecasting, scheduling, flow analysis — into reusable, well-documented building blocks, so the next problem starts from a toolkit instead of a blank file.