Watershed · Engineering profile

Make agent capabilities production-grade.

Mohamed A M Elansary, PhD — production agent evaluation and harness work, observability-minded failure analysis, and climate-adjacent context for sustainability data products.

Agent evals and harnessesObservability-minded failure analysisClimate-adjacent domain contextPython systems

Production agent systems

  • Builds production GPT, Claude, and Gemini agent workflows with retrieval, routing, tenant isolation, provenance, and regression evaluation sets at Vertexium.
  • Treats agent failures as measurable system problems: expected-behavior checks, failure taxonomies, and regression harnesses.
  • Ships Python, API, Bash, and multi-tenant data paths; JavaScript used in production tooling.

Climate-adjacent context

  • Environmental Engineering PhD on HPC multi-basin surface-water and groundwater forecasts with uncertainty quantification over USGS, NOAA, and NASA data.
  • EDAT sustainability and Fortune 500 environmental compliance reporting, including sustainability-reporting frameworks.
  • That context helps when an AI platform sits on climate and ESG operational data. It is not a climate-AI publication trail.

Proposed first contribution

For one Watershed agent capability that touches sustainability data, start by defining the intended user-visible decision, the allowed autonomous vs deterministic vs human-oversight boundary, and a small failure taxonomy. Stand up an evaluation harness with provenance on agent traces, compare simple baselines, attach uncertainty and slice-level failure rates, and make regressions visible before the behavior ships more widely. This is a proposed first contribution shape, not a claim of prior Watershed platform ownership.

Honest fit boundary

This is a Senior Software Engineer AI-platform role that asks for production TypeScript systems and Staff/Sr platform depth. TypeScript and full-stack product-engineering depth are a stretch. I have not owned Watershed-scale agent orchestration and do not invent climate-AI publications. The credible contribution is production agent evaluation and harness work, observability-minded failure analysis, and climate-adjacent domain context.

Role and location

Senior software engineer, AI platform · San Francisco · Ashby workplaceType OnSite. The posting states: “Must be willing to work from an office 4 days per week (except for remote roles)” and “Where we have offices, employees are expected to be in office for 4 days per week.” Willing to relocate to San Francisco with a relocation package. Remote eligibility is not asserted.

Ashby compensation display “$202K – $275K”. Built In display “202K-275K Annually”. · Official role posting