Applied · data science
Messy data in, reliable output shipped.
For focused engagements. This track speaks to turnaround, reproducibility, and clear communication — not research depth. [Tighten to actual offer.]
What I'll do
06applied
Clean & structure your data
Schema enforcement, validation, logging. Traceable inputs, repeatable runs.
07applied
Engineer the features
Rates, lags, rolling windows, spatio-temporal structure from raw counts.
08applied
Forecast with honest error bars
Short-horizon forecasting validated properly — no random splits on time series.
Deliberate non-edges — this funnel does not route to
the deep flagship (02) or the agent-based simulation (09). A gig client doesn't need them and they muddy the "can you ship me something reliable?" read. Keeping the audiences separate. Is this wise?