I built a personal Twitter agent that actually sounds like me. It uses RAG on my own tweets, pulls real context from past lessons, and runs locally (Qdrant + Ollama + Gemma). Replies aren’t generic — they’re grounded, opinionated, earned. This is how you scale signal without losing your voice. Building it in public. Demo ↓
- 0 replies
- 0 recasts
- 0 reactions
weekend build: retrieves my relevant tweets related to given query used stack: 📌 embeddings : nomic-embed-text:latest (ollama) 📌 vectordb: qdrant 📌 lang: python next step: personalised content engine
- 0 replies
- 0 recasts
- 0 reactions
I started working on web3 6 Million ethereum blocks ago Damn!! I feel OLDDD😂
- 0 replies
- 0 recasts
- 0 reactions