Skip to content
View ChaddBrenner's full-sized avatar

Highlights

  • Pro

Block or report ChaddBrenner

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
ChaddBrenner/README.md

Hi, I'm Chadd

Software engineer building data systems and practical automation. I like taking messy operational problems and shipping the system that actually runs them.

Right now I'm the only engineer at a small real estate operator, where I own the whole stack — schema design, ingestion, deployment, backups, and the support call when production breaks. Before that: two summers building Node/GraphQL backend services at a Georgia Tech CREATE-X startup, and two years leading a wearable-sensor machine learning research project at Elon.

Things I've built

Tail of the Dragon Classifier · live demo Eight-class vehicle classifier reaching 99.12% on 23,980 held-out photographs. I screened seven architectures before training ConvNeXt Tiny, then read all 210 remaining misses and wrote up why I stopped instead of chasing another point.

ElonMealsDB · live demo Dining data platform: Python scraper, normalized MySQL schema, read-only Express API with parameterized queries, React frontend. Dockerized, with typecheck, tests, build, and health checks verified on every change.

Wearable Sensor Research About a million raw accelerometer rows across 30 participants. 500+ TensorFlow experiments tracked in MLflow, 96.3% accuracy on step detection, evaluated with subject-independent Leave-One-Group-Out so the numbers reflect people the model never saw. Manuscript submitted to IEEE EMBS BHI 2025.

Tiny NLA Lab Laptop-sized reproduction of the core loop from Anthropic's Natural Language Autoencoders work — hidden activation to English explanation and back to reconstructed activation. Runs locally on Apple Silicon with a 135M-parameter model.

AI Resume Tailor LaTeX resume pipeline driven by a structured facts file, with reviewer agents that flag unsupported claims, layout breaks, and formulaic wording before anything gets sent.

Eufy Lock Codes Local MCP server for smart-lock access codes across rental properties. Previews every change before it runs, expiring execution tokens, redacted audit logs, verification after.


Mostly Python, SQL, and TypeScript. I write things up at chadd.blog.

Portfolio · LinkedIn · Email

Pinned Loading

  1. ElonMealsDB ElonMealsDB Public

    Full-stack dining data system with Python ingestion, a normalized MySQL model, a read-only Express API, and React.

    TypeScript

  2. tail-of-dragon-car-classifier tail-of-dragon-car-classifier Public

    Eight-class ConvNeXt vehicle classifier for Tail of the Dragon road photography, with a full confusion matrix, manual review of every miss, and a React gallery for inspecting predictions.

    Python

  3. mattfeld_research mattfeld_research Public

    Wearable-sensor research pipeline for classifying steps and shifts across 30 participant folds.

    Jupyter Notebook

  4. tiny-nla-lab tiny-nla-lab Public

    Local PyTorch lab for activation reconstruction, probe-backed verbalization, and reproducible prompt experiments.

    Python

  5. ai-resume-tailor ai-resume-tailor Public

    Codex-first resume tailoring workflow with structured facts, LaTeX validation, reviewer agents, and evidence audits

    Shell

  6. eufy-lock-codes eufy-lock-codes Public

    Safety-first Node.js tooling for reviewable, verifiable smart-lock access-code operations.

    JavaScript