Erick Shepherd

← All skills

Skill

Model evaluation

Related work and learning, in one place.

Experience

  • Research Assistant

    UMBC — Department of Physics, ACROS Group · Aug 2017 – Jun 2020

    Atmospheric remote-sensing research on UMBC’s Taki supercomputer. NSF CyberTraining research in fall 2019 and coursework in spring 2020 were part of this longer appointment, with stipend support.

Projects

  • Earth-Science Data Tooling — Packaged Scientific Python

    Project

    I developed scientific-data tools through NASA internships and UMBC research, including conversion to netCDF4 and parsing into pandas DataFrames. cosmic-crunch and noaa-gml-file-reader are published on PyPI; I also contributed to collaborative cloud-anomaly analysis.

  • Gestalt — A Multi-Provider AI Workspace

    Project

    I architect and direct Gestalt’s AI-assisted development: a workspace for conversations, prompt skills, and provider execution. Its Python, React, and Rust components separate prompt content from permissions and connect to tools through MCP.

  • Coding-Agent Evaluation Harness

    Project

    I developed an AI-assisted Python harness to compare coding-agent results against acceptance checks fixed before each run. Recorded provider runs and test results support repeatable comparisons on the selected tasks.

  • LLM Guidance Routing and Evaluation

    Project

    I developed a deterministic Python router for prompt guidance with AI-assisted implementation. It resolves dependencies and context budgets; controlled comparisons tested whether structured revision improved results over the baseline.

Courses & credentials