AI Engineer Roadmap
1–2 weeks

Capstone — Full AI SaaS Product

Build and launch a complete, monetised AI SaaS product targeted at a specific niche.

0%
1–2 weeks0/5 topics done0/14 tasks done

Topics

What to learn
  • Solves a specific, named problem for a specific, named type of user
  • Has a real product name, domain (can be free subdomain), and landing page
  • Landing page converts — someone should understand and want to sign up within 10 seconds
What to learn
  • Stack: Next.js App Router, MongoDB, Clerk auth, Stripe payments, Vercel deployment
  • AI: Anthropic or OpenAI API via Vercel AI SDK, streaming responses
  • Custom system prompt that defines the product's behaviour — AI must feel specialised
  • At least one tool call connecting the AI to your app's data
  • File handling: users can upload at least one file type (PDF, CSV, or plain text)
  • Persistence: conversations and outputs saved per user in MongoDB
  • Rate limiting + Security: no API keys in browser, auth checked on every protected route
What to learn
  • At least 2 pricing tiers — free/freemium + one paid tier
  • Stripe checkout working end to end
  • Webhook updates the user's plan in MongoDB on successful payment
  • Dashboard showing user their usage, current plan, and an upgrade CTA
What to learn
  • DevDocs AI — paste a GitHub repo URL, app fetches README & key files, AI answers questions about the codebase. Target: developers onboarding to new projects.
  • ContractLens — upload a PDF contract, AI highlights risky clauses in plain English. Disclaimer: not legal advice. Target: freelancers and small business owners.
  • DataChat — upload a CSV, ask questions in plain English, AI writes and explains the query, shows results. Target: non-technical people who need to analyse data.
  • JobDescriptionWriter — paste a list of requirements, AI writes a professional JD. Target: startup founders and HR at small companies.
What to learn
  • Write a README (500-800 words) covering what the product does and who it's for
  • Tech stack decisions — why Next.js, why Clerk, why Anthropic vs OpenAI
  • The system prompt you wrote and why you wrote it that way
  • One thing that was harder than expected and how you solved it
  • What you'd build next if you had another week

Assignment

Assignment

Evaluation Checklist

0%0/14

This is the exam. It uses everything from Parts 1–4. This is what you show in job interviews and to potential customers.

Acceptance Criteria

Notes

My Notes
Saved