Skip to content
View brandon-behring's full-sized avatar
3 followers · 0 following

Achievements

Achievement: YOLOAchievement: Pair Extraordinairex4Achievement: Pull Sharkx3Achievement: Quickdraw

Achievements

Achievement: YOLOAchievement: Pair Extraordinairex4Achievement: Pull Sharkx3Achievement: Quickdraw

Block or report brandon-behring

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
brandon-behring/README.md

Brandon Behring

Applied mathematician (PhD). I build systems that are honest about what they don't know.

I work on LLM evaluation and document AI: evaluation harnesses with real statistics, golden datasets with documented rubrics, and extraction pipelines whose accuracy is measured, not asserted. Four years shipping production models at a Fortune-100 insurer; most recently a document-AI research contract. Currently available for evaluation and document-AI work.

Where to look

  • eval-toolkit — evaluation library on PyPI: bootstrap confidence intervals, leakage checks, versioned result schemas, CI gates.
  • prompt-injection-detection-prototype — a full detection study published with its negative result: confidence intervals, baseline contamination disclosed, failure modes analyzed. Honest evaluation, demonstrated.
  • ir-eval — statistical retrieval evaluation for CI/CD: paired tests and drift detection, because most RAG systems fail silently.
  • research-kb — a multi-thousand-source research knowledge base: PDF ingestion → hybrid retrieval (BM25 + vectors + reranking) → MCP server, with a retrieval eval suite gating the weekly rebuild.
  • temporalcv — released Python package for time-series cross-validation with gap enforcement (the leakage everyone ships).

Site: brandon-behring.dev

Pinned Loading

  1. research-kb research-kb Public

    Hybrid-search research knowledge base: PDF ingestion → BM25+vector+rerank retrieval → MCP server, with a retrieval eval suite gating the weekly rebuild

    Python 2

  2. prompt-injection-detection-prototype prompt-injection-detection-prototype Public

    Methodology case study: prompt-injection detector evaluation under cross-family OOD shift. Detectors trained on direct injection meet attack families they did not see — do they still work?

    Python

  3. eval-toolkit eval-toolkit Public

    Reusable binary-classification evaluation toolkit (extracted from prompt_injection_detector PoC).

    Python

  4. research_toolkit research_toolkit Public

    Skill collection for systematic research workflows in Claude Code

    Python

  5. runpod-deploy runpod-deploy Public

    Config-driven RunPod orchestration for reusable GPU project deployments.

    Python

  6. temporalcv temporalcv Public

    Temporal cross-validation for time series

    Python