Mohamed A M Elansary, PhD
Target: Machine Learning Research Engineer, Model Evaluation — WindBorne Systems
Sourced insights
- Eval > headline accuracy: Performance varies across regions, lead times, weather regimes, and customer use cases; standard metrics often fail to capture what makes a forecast meteorologically sound. WindBorne wants scientific taste to decide where models excel, fail, and which results to trust. Source: careers … model-evaluation
- WM-6 public bar: Multi-month eval window (Jul 2025–Mar 2026); up to 38% lower ensemble-mean RMSE vs IFS / 32% vs AIFS; CRPS for skill and ensemble calibration across lead times — multi-metric honesty, not one number. Source: windbornesystems.com/blog/introducing-wm-6
Proof — hard eval × UQ skepticism × agentic tooling × ship
- PhD Environmental Engineering, TAMUK 2022: multimodel / ensemble surface-water/groundwater forecast uncertainty quantification & reduction on HPC — fair baselines, imperfect ground truth, skill before claims.
- Meteorological-skepticism shape: regions, regimes, lead-time horizons (AMS multimodel streamflow; AMS 2021 floods & droughts) — same failure modes WeatherMesh eval must surface.
- Agentic tooling muscle: Vertexium production agentic LLM + retrieval + multi-tenant agents — investigate and synthesize complex outputs; harden recurring analyses into reusable systems.
- Geospatial + time-series + USGS/NOAA/NASA multi-source QA — large scientific datasets under publication-grade validation.
- Honest frame: not claimed WeatherMesh scorecard owner. The PhD who ships. Bay Area OK · Prefer take-home · $140k–$240k + equity story.