| Soseeks Academy — Gen AI Bootcamp
Home Unlock

Gen AI Full Stack Bootcamp — Build Your Own ChatGPT

Build Your Own ChatGPT in 4 Hours 🚀

Hands-on: upload a real PDF, build a vector index with local embeddings, then chat in React while answers come from Groq — fast Llama / Mixtral-class models using a free API key (no credit card on the free tier).

3–4 Hours₹299 · Offline at college

Get Access Code (Pay Offline ₹299)
Stack in class
  • React — upload + chat → your FastAPI
  • FastAPI — PDF → chunks → FAISS
  • sentence-transformers — embeddings on your laptop (no extra API)
  • Groq — ChatGroq + Llama / Mixtral (see model list)

Groq — free API key (~5 min)

Signing up and creating an API key is free for the developer tier (rate limits apply). You do not need a credit card just to get a key for this bootcamp — skip any optional “add payment method” unless you personally want paid limits later.

  1. groq.com → Sign up / log in (email or Google/GitHub if offered).
  2. console.groq.com/keys → Create API key → name it (e.g. soseeks-bootcamp) → copy the key once (gsk_…).
  3. In class you will run: export GROQ_API_KEY="gsk_…" (Mac/Linux) or set GROQ_API_KEY=gsk_… (Windows CMD) in the same terminal as uvicorn.
  4. Pick a chat model id from Supported Models — we default to llama-3.1-8b-instant.

Full click-by-click + optional curl test → see Section 3 on the unlocked bootcamp page.

Session timeline (typical)

BlockFocus
0:00 – 0:45RAG + Groq + local embeddings — mental map
0:45 – 1:45Python venv, Groq key, main.py, Swagger /docs
1:45 – 2:45React + axios, CORS
2:45 – 3:30Try another id from Supported Models (same key)
3:30 – 4:00Q&A, optional SQLite metadata

Bring before class

  • Laptop with 8 GB+ RAM (local embedding model + browser)
  • Python 3.10+, Node.js LTS
  • Groq account + API key from console.groq.com/keys
  • Wi‑Fi for first-time pip / Hugging Face model download
  • A short PDF with real selectable text (not scan-only)

After you unlock

The lab page is a full walkthrough: Groq setup, pip lines, complete main.py (ChatGroq + HuggingFaceEmbeddings), React App.js, request/response tables, and troubleshooting.

Where Gen AI and “database” sit

Gen AI = your backend sends retrieved PDF text + the user question to Groq (Llama / Mixtral). FAISS = vector “database” in RAM built from local embeddings. SQLite is optional for filenames / history only.

Founder intro

Hi, I’m Sanjay, founder of Soseeks.

I’ve built real-world products like Chat2Bill, and this bootcamp is designed to help you understand how ChatGPT works — by building one yourself.

In just 3–4 hours, you will create your own AI chatbot using Python, FastAPI, and React.

No fluff. Only practical learning.

You will ship

  • RAG API with Groq chat
  • React upload + Q&A
  • Clear split: local embeds vs cloud LLM

Pricing — ₹299 (offline)

  1. Pay in college
  2. Get access code
  3. Enter code on website
  4. Open the bootcamp lab page
Enter access code