Inhuman Science

Automated AI/ML content curation and publishing pipeline. Aggregates papers, blog posts, and tweets from the AI world, evaluates them with LLMs, and publishes the best finds to Telegram and Twitter/X.

How It Works

The system runs three independent pipelines on a configurable cron schedule:

Papers — scrapes trending papers from AlphaXiv, scores them with an LLM oracle, downloads PDFs, extracts the most representative figure using a vision model, generates bilingual posts (Russian for Telegram, English for Twitter), and publishes.

Blogs — fetches RSS feeds from OpenAI, Anthropic, and Google Gemini blogs, scores and fact-checks each post, generates summaries, and publishes.

Twitter — monitors tweets from configurable AI leaders (e.g. Sam Altman, Yann LeCun), scores them, generates Russian summaries for Telegram, and retweets on Twitter.

Every pipeline checks for duplicates against recently published content, tracks all decisions in SQLite, and sends error notifications to a dedicated Telegram chat.

Architecture

Sources                 Processing              Publishing
┌─────────────┐        ┌──────────────┐        ┌───────────┐
│  AlphaXiv   │───┐    │  Oracle      │        │ Telegram  │
│  (papers)   │   │    │  (scoring,   │   ┌───▶│ (RU post) │
├─────────────┤   │    │  fact-check, │   │    ├───────────┤
│  RSS Feeds  │───┼───▶│  dedup)      │───┤    │ Twitter   │
│  (blogs)    │   │    ├──────────────┤   └───▶│ (EN post) │
├─────────────┤   │    │  Processors  │        └───────────┘
│  Twitter    │───┘    │  (PDF, image,│              │
│  (tweets)   │        │  post gen)   │              ▼
└─────────────┘        └──────────────┘        ┌───────────┐
                                               │  SQLite   │
                                               │  (state)  │
                                               └───────────┘

Project Structure

├── main.py                 # Entry point, scheduler, pipeline orchestration
├── config.py               # Configuration and environment variables
├── requirements.txt        # Python dependencies
├── Dockerfile
├── docker-compose.yml
│
├── sources/
│   ├── base.py             # ContentItem dataclass
│   ├── alphaxiv.py         # AlphaXiv trending papers scraper
│   ├── blogs.py            # RSS feed parser (OpenAI, Anthropic, Google)
│   └── twitter_feed.py     # Twitter API v2 feed reader
│
├── oracle/
│   └── oracle.py           # LLM-based scoring, fact-checking, deduplication
│
├── processors/
│   ├── pdf.py              # PDF download and text extraction
│   ├── images.py           # Best figure extraction via vision model
│   └── post_generator.py   # Bilingual post generation (RU/EN)
│
├── publishers/
│   ├── telegram.py         # Telegram channel publisher
│   └── twitter.py          # Twitter/X publisher
│
├── storage/
│   └── state.py            # SQLite state tracking
│
└── llm/
    └── client.py           # OpenRouter API client

Setup

Prerequisites

• Python 3.11+
• API keys: OpenRouter, Telegram Bot, Twitter/X

Installation

git clone <repo-url>
cd inhuman-science
pip install -r requirements.txt

Configuration

Copy the example environment file and fill in your credentials:

cp .env.example .env

Required variables

Variable	Description
OPENROUTER_API_KEY	OpenRouter API key for LLM calls
TELEGRAM_BOT_TOKEN	Telegram bot token
TELEGRAM_CHANNEL_ID	Target Telegram channel ID
TELEGRAM_ERROR_CHAT_ID	Chat ID for error notifications
TWITTER_API_KEY	Twitter API key
TWITTER_API_SECRET	Twitter API secret
TWITTER_ACCESS_TOKEN	Twitter access token
TWITTER_ACCESS_SECRET	Twitter access secret

Optional variables

Variable	Default	Description
SCHEDULE_PAPERS_CRON	0 10 * * *	Papers pipeline schedule
SCHEDULE_BLOGS_CRON	0 12 * * *	Blogs pipeline schedule
SCHEDULE_TWITTER_CRON	0 14 * * *	Twitter pipeline schedule
TWITTER_MONITOR_USERS	sama,ylecun,kaborov	Comma-separated Twitter usernames to monitor
ORACLE_MIN_SCORE	7	Minimum LLM score (1-10) to publish
ORACLE_MAX_PAPERS_PER_RUN	5	Max papers published per run
ORACLE_MAX_BLOGS_PER_RUN	3	Max blog posts published per run
TIMEZONE	Europe/Moscow	Timezone for scheduling
DB_PATH	state.db	SQLite database path
PDF_DIR	pdfs	Directory for downloaded PDFs
IMG_DIR	images	Directory for extracted images

Usage

Run a single pipeline

python main.py papers     # Papers pipeline
python main.py blogs      # Blogs pipeline
python main.py twitter    # Twitter pipeline
python main.py all        # All pipelines sequentially

Run the scheduler

python main.py

Without arguments the app starts a background scheduler that triggers each pipeline at its configured cron time and keeps running indefinitely.

Docker

docker-compose up -d

The compose file mounts state.db, pdfs/, and images/ as volumes so state persists across container restarts.

LLM Models

All LLM calls go through OpenRouter. Models are configured in config.py:

Task	Model
Content scoring	deepseek/deepseek-chat-v3-0324
Fact-checking	deepseek/deepseek-chat-v3-0324
Post generation (RU & EN)	anthropic/claude-sonnet-4.6
Figure extraction (vision)	google/gemini-2.5-flash

Data Storage

SQLite database (state.db) with four tables:

• posted_papers — published papers (arxiv ID, title, timestamp)
• posted_blogs — published blog posts (URL, title, timestamp)
• posted_tweets — published tweets (tweet ID, author, timestamp)
• oracle_decisions — all scoring decisions with scores and reasoning