Skip to content
View Tuminha's full-sized avatar
🦷
Building ML benchmarks for dentistry
🦷
Building ML benchmarks for dentistry
7 followers · 23 following

Achievements

Achievement: Pair Extraordinairex4Achievement: Pull Sharkx3Achievement: QuickdrawAchievement: YOLO

Achievements

Achievement: Pair Extraordinairex4Achievement: Pull Sharkx3Achievement: QuickdrawAchievement: YOLO

Highlights

  • Pro

Block or report Tuminha

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
Tuminha/README.md

Francisco Teixeira Barbosa — @Tuminha

Periodontist and builder. Founder of Periospot (14k+ newsletter for dentists) and Executive Director at the Foundation for Oral Rehabilitation. Based in Basel, Switzerland.

I build AI tools that move dentistry from "we think" to "the evidence shows."

What I'm building

  • brar-periodontal-radiograph-benchmark — Leakage-aware ML benchmark predicting periodontitis severity from panoramic radiographs. Negative controls, calibrated baselines, bootstrap model comparisons — built for publication.
  • dental-ai-skills — Structured Claude skills for research critique, GRADE-scored evidence review, and dental content. Works in Claude Desktop, Claude Code, and ChatGPT.
  • llm-evaluation-for-dentistry — Benchmarking LLaMA 3.1 70B providers on dental knowledge. RAG eval with Weights & Biases tracking.
  • dental-evidence-triage — Multi-label classifier that triages PubMed dental abstracts by study design.
  • Periospot-ML-Social-Media-Performance-Predictor — ML model that predicts high-performing posts from pre-publishing features.
  • screenshot-blur-agent — AI-vision pipeline that finds sensitive regions in screenshots and blurs them automatically. Built for Claude Code, Codex, and other agent workflows.

Stack I actually ship with

Python · PyTorch · scikit-learn · Claude API · Next.js · Supabase · Vercel

Find me

periospot.com · LinkedIn · Instagram / @tuminha_dds · YouTube · X / @Cisco_research · cisco@periospot.com

Best way to reach me: email for substantive asks, LinkedIn for quick intros.

Pinned Loading

  1. dental-ai-skills dental-ai-skills Public

    AI Skills for Dentists - Research critique, clinical evidence review, and content creation tools for Claude Desktop and Codex

    Python 4

  2. llm-evaluation-for-dentistry llm-evaluation-for-dentistry Public

    Which LLMs can a dentist trust? A reproducible benchmark of language models on clinical dental knowledge, with clinician-verified rubrics across 6 domains. Part of Periospot.

    Python 9 1

  3. dental-evidence-triage dental-evidence-triage Public

    Multi-label study design classifier for PubMed dental abstracts, enabling triage by automatically categorizing dental evidence according to study design (e.g., randomized trials, observational stud…

    Jupyter Notebook

  4. Periospot-ML-Social-Media-Performance-Predictor Periospot-ML-Social-Media-Performance-Predictor Public

    Machine learning model that predicts high-performing social media posts using pre-publishing metrics and content attributes to optimize content strategy.

    Jupyter Notebook 1

  5. screenshot-blur-agent screenshot-blur-agent Public

    Take screenshots of desktop/Electron apps and automatically blur sensitive information using AI vision + ImageMagick

    Shell 1

  6. brar-periodontal-radiograph-benchmark brar-periodontal-radiograph-benchmark Public

    ML benchmark predicting periodontitis severity from BRAR panoramic radiographs — leakage-audited splits, negative controls, calibrated image baselines

    Python