Zero to a deployed
AI product.
Thirty-one weeks.
No prior experience.
A 31-week programme for 12 selected students — six men, six women — built around one idea: you learn by building. By the end, you have a live product, a real GitHub history, and a skill set matched to what India's fastest-growing job market is actually hiring for right now.
You pay ₹2 lakh only after you're placed — via EMI, over time. If you don't get placed, you owe nothing.
Why now.
India has over 400,000 active AI job openings. By 2027, the country needs 1.25 million AI professionals. Trained supply covers less than 20% of that gap. AI-linked hiring grew 32% in a single year. Entry-level agentic AI engineers earn ₹8–12L starting — above the general AI average — because the specific skills matter.
73% of employers now say they prioritise demonstrated ability and proof of work over formal qualifications. The shift from credential to evidence is already employer behaviour, not a trend article. codeschool's curriculum was verified against real 2026 job postings for Agentic AI Engineer, LLM Engineer, and GenAI Engineer roles before being finalised — close to verbatim what these job postings ask for.
Twelve seats. Not twelve hundred.
Super Builder is deliberately small. Twelve students — six men, six women, by design. Not a batch, not a class. A cohort. You will know every person in it. Your instructors will know you. The retreat, the immersions, the build reviews — everything is designed around twelve people, not scaled for a hundred.
Cohort 1 draws primarily from Kalimpong and Kurseong. Not because people from elsewhere can't apply — but because we're building local proof first, before building wider.
No prior coding required.
Zero. We start from the very beginning. Week 1, Lesson 1: how a computer works.
Commitment over background.
You don't need to have coded before. You need to be the kind of person who finishes things, even when they get hard.
Age 18–28 (a guide, not a wall).
Designed for people at the start of their career who want to build a real foundation.
From the hills, for now.
Kalimpong and Kurseong first. This will expand. It starts here.
How it's structured.
Content Sprint
10–12 live hours every weekend. Every second weekend free. You go from zero to a live, deployed, full-stack AI product. All new concepts are taught in this phase.
Build Arc
~4 live hours every weekend. One product, built continuously. No new concepts. One idea, one product, sixteen weeks. The Retreat sits at Week 14.
Launch
Demo Day prep, Demo Day, placement.
The Content Sprint.
10–12 hours of live class every weekend — Saturday and Sunday, 5–6 hours each day. Every second weekend is free: no new content, just space to consolidate. You arrive not knowing how to code. You leave with a live, deployed, full-stack AI product — and your own product idea locked and mentor-reviewed, ready to build.
Thinking like a computer + JavaScript basics
- How computers actually work
Understand what a CPU, memory, and storage do — so code stops feeling like magic.
- The command line, not the mouse
Navigate folders and files by typing — the daily habit every engineer relies on.
- What a program actually is
See code as a list of instructions a computer follows literally and in order.
- Installing your toolkit + npm basics
Get Node.js, VS Code, and Git installed — and learn npm, how JavaScript projects install external code.
- Your first program: Hello, World
Write, run, and break your first program — and read an error message without panic.
Variables, data, decisions, loops
- Variables — storing information
How a program remembers things while it runs.
- Data types & doing math in code
The basic units of information and how to calculate with them.
- If/else — making decisions
Teach a program to behave differently depending on conditions.
- Lists, loops, and dictionaries
Hold collections of data and repeat work automatically.
- Practice: build a calculator + to-do list
Apply variables, math, decisions, and loops in two small real programs.
Functions, files, error handling
- Functions — reusable blocks of logic
Package logic into named, reusable pieces instead of repeating yourself.
- Working with files
Read from and write to files — how programs save and load real data.
- Handling errors gracefully
Stop programs from crashing when something unexpected happens.
- Intro to object-oriented thinking
Classes and objects as a way to model real-world things in code.
- Practice: build & refactor a budgeting tool
Functions, files, and error handling into one real program.
Git, GitHub, how the web works
- Why version control exists + Git + branches
Save snapshots of your code with clear history, and work on new ideas safely.
- GitHub: your code's home online
Push code to the cloud — repositories are the engineer's real portfolio.
- Client, server, and the internet, plain English
What actually happens when you visit a website.
- HTML & CSS — structure and style
Build the skeleton and look of a real web page.
- Practice: ship a static personal page on GitHub
Your first real, visible, version-controlled thing on the internet.
JavaScript and interactivity
- JavaScript basics + async/await
JS in the browser, and how it handles things that take time without freezing the page.
- Variables, functions, logic in JS
Apply logical thinking to a new context: the browser.
- The DOM — changing a page after it loads
Make a page react to clicks, typing, and other user actions.
- Forms, user input, and events
Build real forms that collect input and respond to what a person does.
- Practice: build an interactive tip calculator
Combine HTML, CSS, and JS into a real interactive tool.
React — component thinking
- Why real websites aren't just one HTML file
The problem component-based frontend tools solve.
- Intro to React: components as building blocks
See a UI as small, reusable, named pieces.
- Props — passing data into components
Make components flexible by feeding them different information.
- State — components that remember things
Let a piece of the UI change itself in response to user actions.
- Browser dev tools + rebuild the tip calculator in React
Inspect, debug, and understand what your code is doing.
Your first backend (Supabase)
- Why you don't need to write a server from scratch
What a managed backend platform does for you.
- Intro to Supabase: database, auth, APIs in one place
Set up a real backend platform without writing server code.
- Designing tables and relationships, visually
Plan real data tables and how they relate to each other.
- CRUD — create, read, update, delete
The four basic operations every application performs on data.
- Practice: build a data-backed app, zero backend code
Wire a real React app to a real, live database.
Auth, APIs, and the wider internet
- Supabase Auth: sign up, log in, sessions
Real user accounts using a trusted, managed service.
- Row-level security & environment variables
Control data access by user and keep secrets out of code.
- REST APIs — methods, status codes, requests & responses
The common language every API speaks.
- Calling a public API from your app
Fetch real, live external data into your own project.
- Practice: add login + combine with an external API
Wire real auth and a third-party API into your project.
Meeting AI — prompting and the LLM API
- What a large language model actually does
A working mental model of how AI models like Claude generate responses.
- Calling Claude's API from your own code
Your first programmatic AI call.
- Prompt engineering fundamentals
Structure, clarity, and specificity for reliable AI output.
- Few-shot prompting and examples
Improve output reliability by showing the model what you want.
- Practice: build an AI feature inside your app
Add a real AI-powered feature to a project you've built.
Deployment + locking your product
- Claude Code as a daily engineering tool
AI-assisted coding as a real workflow — you understand React deeply enough to direct it.
- Deploying your frontend (Vercel) + backend (Supabase)
Get your full-stack app live on the internet, end to end.
- ★ Milestone: your first live, full-stack AI product
Database, auth, frontend, AI feature — all live, deployed, real.
- Choosing your product — idea + architecture
Scope the one product you'll build for the rest of the programme.
- ★ Milestone: lock your idea, begin the Build Arc
Mentor-reviewed plan in hand.
The Build Arc.
No new concepts from here. One product idea, chosen and locked. Sixteen weeks of building it — the same product getting more capable, more stable, more real every week. Support continues every weekend, roughly 4 hours — code review, unblocking, and architecture decisions specific to what each student is building.
Architecture & first build sprint
- Architecture planning + code organization
Diagram your data model, key flows, and major components — structure files so the project stays manageable.
- Build sprint: core data models
Start building the real backbone of your product.
Real APIs with Express + core build
- Building real APIs with Express
Move beyond auto-generated APIs — write your own routes, request handling, and business logic.
- Build sprint: backend logic + AI-assisted frontend
Apply Express; build frontend with Claude Code, the way real engineers work.
Integration + no-code & cross-platform tools
- Connecting frontend to backend, for real
Wire your product's pieces into one coherent, working system.
- Office hours: unblock & debug
Live, open troubleshooting on whatever's stuck.
- No-code tools — speed and cross-platform reach
Lovable for fast prototyping and cross-platform builds. A third skill layer on top of real understanding.
★ The 10-Day Retreat — Beast Mode core build
- Days 1–2: Architecture review and planning
Mentor feedback on each product's system design before heavy building.
- Days 3–4: Core build sprint
Build the real data models, screens, and logic each product depends on.
- Day 5: Guest session — scale engineering
A working engineer on what actually changes when real users show up.
- Day 6: Core frontend build sprint
Build the primary interface a real user would interact with.
- Day 7: Guest session — product & founder conversations
People who've built and launched real products.
- Days 8–9: Integration — end to end
Connect every piece into one coherent, working product.
- Day 10: Checkpoint — show what exists
An honest, unpolished look at your core product.
AI feature: design & RAG
- Designing the AI feature your product needs
Scope a specific AI capability, not AI for its own sake.
- RAG basics: embeddings & vector search (pgvector)
Store and query your product's own data for AI to retrieve from.
Agents & tool use + MCP
- Agents and tool use
Give AI the ability to call real code via the Claude Agent SDK.
- MCP — Model Context Protocol
How agents connect to tools and data in 2026. If you're not using MCP, you're building on legacy infrastructure.
Integrating AI into the product
- Connecting AI to the frontend + streaming
Build a real chat or search interface; make AI output feel fast, not silent.
- Cost optimization for AI features
Per-conversation cost, prompt caching, model routing by query complexity.
Evaluation & AI quality
- Eval design — real eval literacy
Offline eval set, LLM-as-judge online eval, tracking task success, latency, and cost per task.
- ★ Milestone: core AI feature live
The AI capability that defines your product, fully working end to end.
Resilience & safety
- Error handling and input validation
Stop bad data from ever reaching your database or your AI calls.
- Safety & guardrails (OWASP LLM Top 10)
Real failure modes including Excessive Agency. Mitigations: least-privilege tool design, human-in-the-loop, kill switches.
Testing & real users
- Writing real tests for your product
Unit and integration tests that catch bugs before users do.
- Getting real users + acting on feedback
Find real people to try your product and turn feedback into changes.
Database performance & security
- Indexes and query performance
The single highest-leverage database performance technique.
- Security audit — thinking like an attacker
Audit auth, data access, and secrets before anyone else finds the gaps.
Scale & multi-agent patterns
- Rate limiting & protecting AI costs
Prevent a single user or bug from running up an unbounded bill.
- Multi-agent orchestration
Coordinate multiple agents or AI steps reliably — only where the product genuinely calls for it.
Frontend polish & UX
- UI polish, responsive design, accessibility basics
Close the gap between functional and something people enjoy using.
- ★ Milestone: a product that feels finished
Frontend polish that makes your product feel real, not just functional.
Shipping discipline
- CI/CD — what's already built for you
The automatic deploy pipeline Vercel and Supabase already give you.
- Final feature push: core gaps
Close the most important remaining gaps against your original plan.
Final hardening
- Monitoring, backups, final security pass
Know when something breaks; plan for the worst case.
- ★ Milestone: production-stable
Feature-complete, secure, monitored, ready to be shown.
Career readiness immersion
- Resume + portfolio around your capstone
Turn months of real work into a document and portfolio that gets noticed.
- Mock interviews — technical + behavioral
Practice explaining your work under real interview pressure.
The Launch.
Demo Day preparation
- Telling your product's story + building the presentation
Craft and structure a compelling walkthrough of what you built and why.
Final polish
- Last UI/copy pass, bug triage, documentation
Fix visible rough edges; document how your system works.
Demo Day rehearsal immersion
- Full run-throughs + presentation coaching
Practice in front of the cohort and mentors before it counts.
Demo Day
- Demo Day: presentations & networking
Present live to guests, employers, and the wider community — codeschool's first public proof point.
Placement & transition
- Follow-up, continued applications, offer negotiation
Turn Demo Day into real outcomes — applications, interviews, and a first offer.
What you learn, month by month.
JavaScript, Node & npm, Git & GitHub, HTML/CSS, VS Code, Command Line
Can write, run, and debug real JavaScript programs from scratch, and ship a version-controlled static page live on GitHub. No prior coding assumed.
JavaScript (DOM, async/await), React, Browser Dev Tools, Supabase, REST APIs
Can build a real, interactive, data-backed web app with user accounts — full front end in React, backend via Supabase.
Claude API, Claude Code, Prompt Engineering, Vercel
Has a live, deployed, full-stack AI product on the internet, and a locked, mentor-reviewed plan for the capstone.
Express.js, system architecture, no-code tools (Lovable-style)
Has architected and started building the real capstone — with a true backend — and pushed its core forward fast during the Retreat.
RAG / pgvector, Claude Agent SDK, MCP, eval design
The product's core AI feature — retrieval, agents, and tool use via MCP — is fully built, evaluated, and working end to end.
Testing frameworks, OWASP LLM Top 10, database indexing, security auditing, accessibility
The product is hardened against real-world misuse, has a real test suite, and has been used by real people.
CI/CD, monitoring, resume & interview prep
The product is feature-complete and production-stable; ready to apply and interview with a real, deployed body of work.
Presentation skills, documentation
The product is demoed publicly at Demo Day, and the placement push is fully underway.
Five days, in the hills.
Five full-day, in-region sessions across the programme — placed at moments where being in a room together genuinely matters.
Kickoff
The cohort meets in person for the first time. Why we're here. What AI-native building looks like. Industry guests join.
Content Sprint Complete
Show your first live product. Industry guests attend. A real milestone, marked in person.
Build Crit Day
Working engineers give real, honest critique of the products being built. Sometimes uncomfortable. Always useful.
Career Readiness
A full day on resume, portfolio, and mock interviews — built around what you actually shipped.
Demo Day Rehearsal
Full presentations in front of the cohort and mentors, before the real thing.
What you'll know cold.
JavaScript / TypeScript
Front end and back end, one language throughout
React
Component-based UI
Express.js
Real backend APIs, your own business logic
Supabase
Database, auth, and instant prototyping
Vercel
Deployment
Claude Agent SDK
Agentic AI, built in from Week 9
MCP
How agents connect to tools and data in 2026
Lovable
No-code and cross-platform, a real third skill layer
Note: Exact tools may shift before launch. AI moves fast and we'll always teach what's current, not what was convenient when this was written.
Three working engineers. Not a rotating cast.
One consistent teaching relationship per phase. Not a new face every week. Each instructor is with you long enough to know your product, your blockers, and where you actually are.
Content Sprint & Full-Stack Lead
3–6 years experience, genuine full-stack range: JavaScript/TypeScript, React, Express, Supabase — plus real hands-on experience with no-code and AI builder tools. The only teacher you'll know for your first three months. The single most important hire for cohort continuity.
AI & Agent Specialist
5–8 years experience, specific hands-on RAG pipelines and agentic AI systems: tool-calling, multi-step reasoning, MCP. The highest-ceiling, most market-critical content in the programme. Someone who has actually done this in production.
Production Engineering Lead
5–8 years experience, senior backend and full-stack: reliability, security, testing, scale, shipping discipline. The stretch most courses skip entirely. Exactly what makes the difference between a demo and a product that survives real users.
Plus — network-sourced, not a paid budget line: product mentors at idea-selection, guest engineers at the Retreat, career speakers at the immersions. Drawn from codeschool's own founder and engineering network.
What you walk out with.
A product, not a certificate.
By the end, you can build a startup's entire product — front end, back end, database, live AI feature — alone, with the fastest stack available. Not a tutorial clone. Something you designed, scoped, built, deployed, and put in front of real users.
Proof in 60 seconds.
Live URL. Real GitHub history. A recruiter can verify your work before they've finished reading your name. Nothing to take on faith.
A line to employment.
Demo Day puts your product in front of people who hire. Your instructors become references. Your cohort goes to market alongside you. Entry-level GenAI engineers with this exact skill set earn ₹8–12L starting — the curriculum was built to match that bar exactly.
You don't need a CV. You'll graduate with one.
Free. Entirely.
No upfront fee. Tuition is free during the programme.
You pay ₹2 lakh only after you're placed — via EMI, over time, once you're earning. If you don't get placed, you owe us nothing.
This is an Income Share Agreement. You only pay if it works.
Twelve seats.
Darjeeling will build.
Cohort 1 starts mid-2027. Kalimpong and Kurseong first. No upfront cost. No risk if you don't place.