llm-pipeline

An agentic extractor for messaging platforms. Noria receives messages & URLs via chat; scrapes web data to evaluate the content against custom scoring matrices (e.g., jobs, scholarships); and sends a structured summary with a calculated match score directly to the assigned messenger number.

Karpathy’s LLM Wiki, 100% local with Ollama. Drop Markdown notes → AI extracts concepts → your Obsidian wiki auto-links and grows. Zero sharing. Your notes stay yours.

More than just Karpathy’s LLM Wiki, 100% local with Ollama. Drop Markdown notes → AI extracts concepts → your Obsidian wiki auto-links and grows. Zero sharing. Your notes stay yours.

Bounded, inspectable LLM inference pipelines from declared YAML — runs offline against Ollama or any OpenAI-compatible local server, emitting JSONL traces, inspect reports, and stable exit codes.

MCP server that shrinks logs, JSON, docs and web pages before your AI reads them. Offline, deterministic, ~3ms, no API key. 66% fewer tokens on a mixed dev corpus, 90%+ on logs.

Agent skills that turn Claude Code (or any agent) into an award-winning multi-modal creative pipeline — art, music, video, story. Panel-of-Experts debate, a deterministic state graph, and an immutable continuity block that makes creative drift structurally impossible. Prose contracts, not framework code.

Simplified Technical English for Code - extract, adapt, and enforce ASD-STE100 rules for software documentation. 9-agent pipeline, 59-test benchmark (96.6%), multi-language scaffolding.

Document Summarization App using large language model (LLM) and Langchain framework. Used a pre-trained T5 model and its tokenizer from Hugging Face Transformers library. Created a summarization pipeline to generate summary using model.

End-to-end Claude Code skill for job applications. 8 phases, 2 sub-skills, originally built for UK AI startup hunt, generalised. Bilingual EN/中文.

Automated 0DTE options scanner + AI signal bot. Scans 50 US tickers, scores 698+ contracts via Delta/Vol-OI/Gamma formula, selects Kill Shot, generates Groq AI explanation, delivers Discord Embed alerts 3x/day. Deployed 24/7 on Railway.app with PostgreSQL logging.

Long-form novel analysis pipeline with chapter extraction, report export, and quality review

A powerful CLI tool for extracting text from documents using DeepSeek OCR and generating high-quality datasets with LLM assistance.

Local-first pipeline for a daily AI news digest

Self-hosted batch LLM pipeline for analyzing customer feedback from Excel. Upload xlsx, describe the task, configure output fields — get structured results. Works with any OpenAI-compatible API and Ollama.

Universal LLM-powered lesson generator for solo learners across code, medicine, and language. Self-contained interactive lesson HTML. Q8 architecture: parent Claude Code agent runs all text-LLM steps; built as a Claude Code skill.

信息高度图名的时代，真正稀缺的不是信息，而是结构化的判断力。每天面对 arXiv、Hacker News、TechCrunch 等渠道的海量信息资源，真正需要的是一个能在信息噪音中提取信号、在碎片中构建体系的洞察引擎。

Trade-Alert is a notification system that keeps users updated on critical news impacting their stock portfolios. It simplifies staying informed by delivering timely notifications for important articles, eliminating the need to monitor multiple platforms in today’s fast-paced market.

无人值守的 B 站直播录播切片流水线，每一步 fail-closed｜Unattended, fail-closed VTuber stream clipping pipeline for Bilibili

OpenSCAD 3D model generation & refinement pipeline with a silly website for comparing open-weight LLMs.

Build a personal LLM wiki using Claude Code with a two-layer cache architecture that integrates with Logseq and Obsidian.