Skip to content
View DerekW00's full-sized avatar
4 followers · 8 following

Achievements

Achievement: Pull Sharkx2Achievement: QuickdrawAchievement: YOLOAchievement: Pair Extraordinaire

Achievements

Achievement: Pull Sharkx2Achievement: QuickdrawAchievement: YOLOAchievement: Pair Extraordinaire

Block or report DerekW00

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.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. 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
DerekW00/README.md

Hi, I'm Derek

Software engineer focused on reliable AI systems, backend infrastructure, and full-stack products.

I'm based in San Francisco. My work spans healthcare and fintech, with a focus on making AI behavior measurable and production failures easier to diagnose.

Previously, I was a founding engineer at Double Blind Bio, working on a clinical-trial AI product, evaluation tooling, observability, and backend reliability. Before that, at Baselayer, I built business-verification services, risk-scoring workflows, and agent-assisted data enrichment.

Start here: LLM Reliability Lab

llm-reliability-lab is an offline Python evaluation harness for structured invoice extraction. It demonstrates how a schema-valid change can corrupt an invoice identifier, how field-level metrics and critical-error gates catch it, and how a baseline comparison verifies the repair.

It uses synthetic documents and authored predictions. It demonstrates evaluation mechanics; it does not report live-model accuracy or production results.

Quickstart · Walkthrough · Example reports

Other projects

  • Photoreceptor Eye Model: Python and OpenCV experiments in afterimage simulation and image/video processing.
  • NLP API: a small FastAPI and Hugging Face prototype exposing sentiment analysis and text-generation endpoints.
  • ADHD Assistant: an archived Next.js and TypeScript prototype for task planning, habits, and focus sessions.

Tools I work with

Python · TypeScript · SQL
FastAPI · Node.js · Next.js · React · PostgreSQL / pgvector
LLM evaluation · LangGraph · DSPy · Docker · GCP · observability

I graduated from UC Berkeley with a B.A. in Computer Science, Data Science, and Economics.

Open to software engineering roles in applied AI, backend, and full-stack product development.

LinkedIn · Portfolio

Pinned Loading

  1. support-operations-desk support-operations-desk Public

    Human-reviewed support operations with cited fixture drafts, deterministic SLA policy and durable PostgreSQL jobs.

    Python

  2. incident-evidence-workspace incident-evidence-workspace Public

    Evidence-first incident investigations with scoped tools, durable jobs and auditable hypotheses.

    Python

  3. invoice-reconciliation-workbench invoice-reconciliation-workbench Public

    Exact-money invoice reconciliation with atomic CSV imports, versioned approvals and frozen ledger exports.

    Python

  4. llm-reliability-lab llm-reliability-lab Public

    Offline invoice-extraction evaluation: strict schemas, field metrics, critical-error gates, and replayable regression checks using synthetic fixtures.

    Python

  5. photoreceptor-eye-model photoreceptor-eye-model Public

    Python + OpenCV experiments in photoreceptor response, afterimage simulation, and image/video processing.

    Python

  6. nlp-api-project nlp-api-project Public

    FastAPI + Hugging Face prototype with sentiment-analysis and text-generation endpoints.

    Python