PrototypeAI/LLMresearch

Multimodal PDF Review Pipeline

Scrapes PDFs, runs a multimodal judge + text critic via local Ollama, emits an HTML report.

Speedup: 7.28×
Mode: Map-reduce judge
Test set: 28 pages · 18.4 min
Backend: Ollama multimodal

Page in preparation

Multimodal PDF Review Pipeline is real and active — the write-up is on the queue.

Stats above are pulled from the working repo, not invented. A full page with premise, architecture, and current state will land here in the next content pass.

Back to selected work

∴⎯Related work

Prototype

LLM Self-Improvement Loop

Rejection-sampling fine-tuning of Qwen3.6:35B-A3B on a single AMD V620, weights-mutating not prompt-tweaking.

AI/LLMtoolingresearch

Production

Open-Loop Daily Digest

An experiment running in public: a local LLM writes a daily research digest with zero human curation, and posts it here.

AI/LLMtoolingresearch

Prototype

Ruin Detection from Orbit

Vision pipeline that hunts for archaeological anomalies in satellite tiles with LLM-assisted confidence scoring.

AI/LLMresearch