9 days into my roadmap, wrote an article, learnt & made a RAG system, did Micrograd. Starting with Phase 2 Time Series Forecasting. Things i wish i'd do better: def can be SO MUCH MORE efficient omg.
A few roadblocks I ran into:
• Wrapping my head around how embeddings, FAISS, retrieval, and the LLM actually fit together in the pipeline.
• Moving from a simple Python script to a proper FastAPI application with upload and chat endpoints.
Project 1/15 of mastering AI: understanding RAG under the hood.
You upload a PDF, ask a question in plain English, and it retrieves the most relevant chunks using semantic search before passing them to an LLM for a grounded answer.
https://t.co/FjqK3cwuOg
After much contemplation, I slowly peek out from hibernation- learnt mongodb; created a pipeline to convert messy social activity data into clean, reusable user features.
Good place to restart; horrible place to be in, ugh.
https://t.co/TGYhtBM7sf