AI-powered Indian news aggregator that cross-references stories across multiple trusted sources before publishing. The whole pipeline runs on its own — fetch, verify, write, publish.
AI Verification Pipeline :-
- Automated Fact-Checking: Every story is verified using Groq Llama 3.3 70B before it reaches your screen.
- Multi-Source Cross-Reference: Stories must be confirmed by 2+ independent trusted sources or they are discarded.
- Credibility Scoring: Every article receives a 0–100 credibility score based on source authority, detail richness, and writing tone
- Neutral AI Writing: Verified facts are rewritten into unbiased, factual summaries — no sensationalism, no spin.
Fully Autonomous :-
- Runs 24/7: Fetches news from RSS feeds and APIs every 30 minutes — no human editors needed.
- SHA256 Deduplication: Duplicate stories are automatically filtered out using URL hashing.
- Domain Blocklist: Known satire, spam, and fake-news domains are blocked at the pipeline level.
- Structured Logging: Full observability with detailed pipeline logs to stdout.
Live - https://verifiedindian.vercel.app/
GitHUB - https://github.com/roshhellwett/projectsentinel.git
Tech Stack :-
Layer Technology
Frontend Next.js 15 + TypeScript + Tailwind CSS
Backend Python 3.11 + FastAPI
Database Supabase (PostgreSQL)
AI Verification Groq API (Llama 3.3 70B)
AI Writing Groq API (Llama 3.3 70B)
Hosting Vercel (frontend), Railway (worker)



Top comments (0)