DEV Community: Ramesh Reddy Adutla

I Built a Free AI Education Platform in 10 Languages — And Open-Sourced Everything

Ramesh Reddy Adutla — Mon, 09 Mar 2026 19:01:11 +0000

I Built a Free AI Education Platform in 10 Languages — And Open-Sourced Everything

AI Educademy is a free, open-source AI education platform I conceived, designed, and built entirely from scratch — every line of code, every lesson, every language.

What it is

🎓 112+ interactive lessons across 3 difficulty tracks (Seeds → Sprouts → Branches)
📝 18 SEO blog posts covering AI fundamentals, career paths, ethics, and more
🌍 10 languages: English, French, Dutch, Hindi, Telugu, Spanish, Portuguese, German, Chinese, Japanese (Arabic coming next)
🔬 AI Lab / Playground — hands-on experiments, no setup needed
⚡ 97 PageSpeed score on desktop, 97 on mobile — zero Framer Motion, CSS-only animations
📖 FAQ page with schema.org markup for Google rich results
📑 Blog table of contents — auto-generated from headings, sticky sidebar on desktop
🔒 100 Best Practices, 100 SEO on Lighthouse
Built with Next.js 15, TypeScript, MDX, Tailwind CSS

Why I built it

AI will reshape every industry. Yet the best AI education remains locked behind expensive institutions in a handful of countries. I've worked across Asia and Europe — in fintech, media, and retail — and it's clear: language should never be a barrier to learning.

The tech stack

Next.js 15 (App Router) + TypeScript
next-intl — i18n routing for 10 locales
MDX — content as code (lessons + blog)
Tailwind CSS — zero runtime CSS-in-JS
Vercel — deployment & analytics
Vitest — 122 unit tests
GitHub Actions — CI on every push

What's next

Arabic (RTL support already built in)
More lessons across all 3 tracks
Spanish, Portuguese, Hindi blog post translations
Community contributions welcome!

Open source

Platform: github.com/ai-educademy/ai-platform
Content: github.com/ai-educademy/ai-courses

If this resonates with you, ⭐ the repo or share it with someone who wants to learn AI in their own language.

Built solo, shipped with ❤️ from London.

I Built a Free, Multilingual AI Education Platform — Here's How 🌍🤖

Ramesh Reddy Adutla — Fri, 06 Mar 2026 14:09:47 +0000

What if anyone in the world could learn AI — for free, in their own language?

That question kept nagging me.

I'm an engineering leader who's spent years building systems at scale. But when I looked at the AI education landscape, I saw the same pattern repeating: great content locked behind paywalls, taught exclusively in English, and assuming you already have a CS degree.

My mother speaks Telugu. My colleague's parents speak Hindi. A friend in Brussels works in Dutch. None of them could find a single, structured AI course in their own language that didn't cost hundreds of dollars.

So I built one.

AI Educademy is a free, open-source platform that teaches AI from absolute zero to production mastery — in 5 languages, with 10 progressive programs, powered by a modern Next.js stack.

🔗 Live: aieducademy.vercel.app
⭐ GitHub: github.com/ai-educademy/ai-platform

The Problem Nobody's Solving

Here's what I found when I surveyed AI education:

Fragmented — YouTube playlists, random blog posts, half-finished Udemy courses. No coherent path from "What is AI?" to "Deploy an ML pipeline."
English-only — 75% of the world doesn't speak English as a first language. If you're a curious teenager in Hyderabad or a career-changer in Lyon, your options are... limited.
Paywalled — The best structured courses cost $50–$500. That's a month's salary in some countries.
Assumes prerequisites — Most courses start with "Let's import TensorFlow" before explaining what a neural network actually is.

AI is going to reshape every industry. The people building it — and the people affected by it — shouldn't need a credit card and fluent English to understand it.

The Solution: AI Educademy 🌱→🌲

AI Educademy uses a nature growth metaphor. You start as a seed and grow into a forest:

🌱 AI Seeds        → "What is AI?" (no code, no maths)
🌿 AI Sprouts      → Foundations (data, algorithms, neural nets)
🌳 AI Branches     → Specialisations (ML, CV, NLP, GenAI)
🏕️ AI Canopy       → Production AI (MLOps, RAG, governance)
🌲 AI Forest       → Mastery (research, leadership, frontier AI)

Plus a parallel craft-engineering track for building real AI products:

✏️ AI Sketch       → Design thinking for AI products
🪨 AI Chisel       → Engineering fundamentals
⚒️ AI Craft        → Building real AI systems
💎 AI Polish       → Production-grade refinement
🏆 AI Masterpiece  → Capstone portfolio projects

10 programs. 2 tracks. 5 languages. 150+ MDX lessons. All free. All open source.

Every lesson is available in English 🇬🇧, French 🇫🇷, Dutch 🇳🇱, Hindi 🇮🇳, and Telugu 🇮🇳 — with a content fallback chain so missing translations gracefully degrade to English.

Architecture Deep Dive 🏗️

This is the part you're here for. Let me walk through the key technical decisions.

Next.js App Router + next-intl

The platform runs on Next.js with the App Router. Every route is locale-aware via next-intl:

src/app/[locale]/
├── page.tsx                              # Homepage
├── programs/[programSlug]/
│   └── lessons/[slug]/page.tsx           # Lesson viewer
├── dashboard/page.tsx                    # Progress tracking
├── playground/page.tsx                   # AI games
└── about/page.tsx

The [locale] segment is handled by next-intl middleware, which detects the browser language, matches it against supported locales (en, fr, nl, hi, te), and routes accordingly. A LanguageSwitcher component lets users override at any time.

MDX Content System

Each lesson is an MDX file with rich frontmatter:

---
title: "What is AI? A Friendly Introduction"
description: "No jargon, no maths — just a simple explanation..."
order: 1
difficulty: "beginner"
duration: 10
icon: "🤖"
published: true
---

## Welcome! 👋

If you've ever heard people talk about "AI" and thought
_"What even is that?"_ — you're in exactly the right place.

<Illustration
  src="/images/lessons/what-is-ai.svg"
  alt="Traditional software follows rules. AI learns patterns."
  caption="The core difference: rules vs learning"
/>

MDX gives us the best of both worlds: authors write in Markdown (low barrier), but we can embed React components like <Illustration>, <Quiz>, and <CodePlayground> directly in lesson content.

We use next-mdx-remote for rendering and gray-matter for frontmatter parsing at build time.

The Content Fallback Chain

This was a key design decision. Not every lesson has been translated into all 5 languages yet. So the content loader implements a fallback:

// Simplified from src/lib/lessons.ts
export function getLesson(program: string, slug: string, locale: string) {
  const localePath = `content/programs/${program}/lessons/${locale}/${slug}.mdx`;
  const enPath     = `content/programs/${program}/lessons/en/${slug}.mdx`;

  // Try locale-specific first, fall back to English
  const filePath = fs.existsSync(localePath) ? localePath : enPath;
  const raw = fs.readFileSync(filePath, "utf-8");
  const { data, content } = matter(raw);
  return { meta: data, content };
}

This means contributors can translate one lesson at a time without breaking anything. The platform always works — it just gracefully degrades to English for untranslated content.

Git Submodules: The Secret Sauce

Here's what makes the architecture scale. Each of the 10 programs is its own GitHub repo, pulled into the platform via git submodules:

ai-platform/content/programs/
├── ai-seeds/          ← github.com/ai-educademy/ai-courses
├── ai-sprouts/        ← github.com/ai-educademy/ai-courses
├── ai-branches/       ← github.com/ai-educademy/ai-courses
├── ai-sketch/         ← github.com/ai-educademy/ai-courses
├── ai-chisel/         ← github.com/ai-educademy/ai-courses
├── ai-craft/          ← github.com/ai-educademy/ai-courses
├── ai-polish/         ← github.com/ai-educademy/ai-courses
├── ai-canopy/         ← github.com/ai-educademy/ai-courses
├── ai-forest/         ← github.com/ai-educademy/ai-courses
└── ai-masterpiece/    ← github.com/ai-educademy/ai-courses

Why submodules over a monorepo?

Independent versioning — Content authors can work on ai-seeds without touching platform code.
Separate permissions — Translators get write access to content repos only.
Auto-deploy — Each content repo has a GitHub Action that fires repository_dispatch to the platform repo, triggering a Vercel rebuild. Content changes go live in ~60 seconds.
Clean separation — The platform is the shell; content is pluggable. Want to add a new program? Create a repo, add a submodule, register it in programs.json. Done.

Shared Design System

All UI components come from @ai-educademy/ai-ui-lib — a separate npm package with its own Storybook. The platform consumes it; the content repos don't need to know about it. Consistent design across everything.

PWA + Offline Support

The platform is a Progressive Web App powered by Serwist (a modern Workbox wrapper for Next.js). Users in areas with spotty internet can install it and access cached lessons offline. For an education platform targeting global learners, this isn't a nice-to-have — it's essential.

CI/CD Pipeline

Push to main (or content repo dispatch)
    ↓
GitHub Actions CI
    ├── ESLint (strict TypeScript)
    ├── Next.js production build
    └── Playwright e2e tests (multi-browser)
    ↓
Vercel auto-deploy → aieducademy.vercel.app

The Content Architecture

Here's how 150+ MDX files are organized across 10 programs × 5 locales:

content/programs/
└── ai-seeds/                    # Each program = one git repo
    ├── lessons/
    │   ├── en/                  # English (canonical source)
    │   │   ├── what-is-ai.mdx
    │   │   ├── how-machines-learn.mdx
    │   │   └── your-first-ai-model.mdx
    │   ├── fr/                  # 🇫🇷 French translations
    │   │   ├── what-is-ai.mdx
    │   │   ├── how-machines-learn.mdx
    │   │   └── your-first-ai-model.mdx
    │   ├── nl/                  # 🇳🇱 Dutch
    │   ├── hi/                  # 🇮🇳 Hindi
    │   └── te/                  # 🇮🇳 Telugu
    ├── assets/                  # Illustrations, diagrams
    └── program.json             # Program metadata

The slug-based naming (what-is-ai.mdx) means the same slug resolves across all locales. The platform's programs.json registry ties everything together:

{
  "programs": [
    { "slug": "ai-seeds", "level": 1, "icon": "🌱", "track": "ai-learning" },
    { "slug": "ai-sprouts", "level": 2, "icon": "🌿", "track": "ai-learning" },
    { "slug": "ai-sketch", "level": 1, "icon": "✏️", "track": "craft-engineering" }
  ]
}

Adding a new program to the platform is literally: add one JSON entry, add one git submodule. The routing, i18n, and lesson rendering all work automatically.

Cool Features ✨

🎮 AI Playground — An interactive drawing recogniser where you sketch objects and a pre-trained model guesses what you drew. It's the "Hello World" of experiencing AI firsthand — no installation, no setup, just draw.

📊 Dashboard — Track progress across all programs. Completion percentages, learning streaks, and certificates for finished programs.

🌙 Dark / Light Mode — System preference detection with a manual toggle. Because we're developers and dark mode is non-negotiable.

🔐 Auth + Guest Mode — GitHub OAuth via NextAuth.js for persistent profiles, but you can also learn as a guest with localStorage-based progress. Zero friction to start.

🎨 Smooth Animations — Framer Motion powers page transitions and micro-interactions throughout the platform. It feels good to use.

Tech Decisions & Trade-offs

Every architecture has trade-offs. Here are mine, honestly:

MDX over a CMS

Why: Content lives in git. Version control, PR-based reviews, no vendor lock-in, and contributors use their existing GitHub workflow.
Trade-off: No WYSIWYG editor. For a developer-oriented open-source project, that's actually a feature, not a bug.

Git submodules over a monorepo

Why: Content and code have different lifecycles, different contributors, and different release cadences.
Trade-off: Submodules have a learning curve. git clone --recurse-submodules isn't intuitive for newcomers. But for independently versioned content plugged into a shell — they're the right tool.

localStorage for the MVP

Why: Zero backend. No database to manage, no server costs, no GDPR headaches. Users start learning in 2 seconds without signing up.
Trade-off: Progress doesn't sync across devices. That's the first thing I'll upgrade — likely with a lightweight backend or edge-based storage.

Vercel for deployment

Why: Generous free tier, edge functions work great with next-intl, and git-push-to-deploy is frictionless.
Trade-off: Vendor dependency. But the app is standard Next.js — porting to Docker or any Node.js host would take an afternoon.

What I Learned 📝

i18n is hard. Multilingual content is really hard. The routing and message files were straightforward with next-intl. The hard part is creating quality translations at scale — culturally appropriate, not just machine-translated.
Content architecture matters more than code architecture. I spent more time designing the MDX structure, frontmatter schema, and fallback chain than the React components. Get the content system right, and everything else follows.
Git submodules have a bad reputation they partly deserve. But for this specific use case — pluggable content repositories with independent CI — they're genuinely the best tool.
PWA support in Next.js has matured beautifully. Serwist made offline support almost trivial. For education platforms targeting users with unreliable internet, this is a game-changer.
Open source is a multiplier. Within weeks, I had contributors offering translations in languages I hadn't even considered. The community effect is real — and humbling.

How You Can Help ⭐

If you've read this far, you probably care about making tech education accessible. Here's how to get involved:

🌍 Contribute a translation

Every lesson translated is another person who can learn AI in their language. Even one lesson helps. The MDX format is just Markdown — you don't need to be a developer.

⭐ Star the repo

Stars drive discoverability on GitHub. More stars → more contributors → more languages → more learners.

📢 Share it

Know someone learning AI? Someone who teaches? Someone who speaks French, Dutch, Hindi, or Telugu? Send them the link.

🛠️ Contribute code

The platform is Next.js + TypeScript + Tailwind. The design system has a Storybook. If you can build React components, you can improve AI Educademy.

🔗 Links:


🚀 Live platform	aieducademy.vercel.app
⭐ GitHub	github.com/ai-educademy/ai-platform
🎨 Storybook	ai-educademy.github.io/ai-ui-lib
📦 npm	@ai-educademy/ai-ui-lib

AI education shouldn't be a privilege. Let's make it a right.

Built with ❤️ by Ramesh Reddy Adutla — engineering leader, open-source advocate, and believer in education without borders.

I Built a Free AI Education Platform for the World — Heres How You Can Help 🌱

Ramesh Reddy Adutla — Wed, 04 Mar 2026 23:33:55 +0000

Last year, I noticed something frustrating: most quality AI education is either in English only, locked behind expensive paywalls, or assumes you already have a CS degree.

As someone from India, I know firsthand how many brilliant students are left behind simply because they don't have access to resources in their language.

So I built AI Educademy — a free, open-source, multilingual platform to learn AI from absolute scratch.

What is AI Educademy?

🌍 5 languages — English, French, Dutch, Hindi, Telugu (more coming!)
🧪 Interactive — AI Playground with TensorFlow.js (draw something, watch AI recognise it!)
📱 Mobile-first — works on any device
💝 Free forever — MIT licensed, no paywalls, no ads
🔓 Open source — built with Next.js 15, TypeScript, TailwindCSS

The Tech Stack

Technology	Purpose
Next.js 15	Framework (App Router, Server Components)
TypeScript	Type safety
TailwindCSS	Styling
next-intl	Internationalisation (i18n)
MDX	Lesson content
TensorFlow.js	In-browser AI demos

How You Can Help

We have good first issues waiting for contributors:

🇯🇵 Add Japanese support
🇸🇦 Add Arabic support (RTL!)
🇩🇪 Add German support
🇪🇸 Add Spanish support
🎨 Add dark mode
📊 Add progress tracking

No AI knowledge needed to translate — just your language skills!

👉 github.com/ai-educademy/ai-platform

🌐 aieducademy.vercel.app

Star ⭐ the repo if you believe AI education should be free for everyone.

I'd love your feedback — what topics should we cover next?

I Built a RAG Bot That Lets You @mention Your Entire Codebase in GitHub Copilot

Ramesh Reddy Adutla — Wed, 04 Mar 2026 21:02:54 +0000

Ever wished you could ask your codebase a question and get an answer that actually cites the source file?

I built GitSage — a self-hosted RAG bot that indexes your entire GitHub org and lets you chat with it. The killer feature? It works as a GitHub Copilot Extension, so you can type @gitsage right inside Copilot Chat.

You:      @gitsage how does authentication work across our services?

GitSage:  Authentication is handled by the auth-service using JWT tokens...
          📁 auth-service/src/main/java/AuthController.java (score: 0.94)
          📁 auth-service/src/main/java/JwtTokenProvider.java (score: 0.89)

No hallucinations. Just your code, cited.

The Problem

I work in a large org with 100+ repositories. New joiners take weeks to understand the codebase. Senior engineers repeatedly answer the same "where is X?" and "how does Y work?" questions.

GitHub Copilot is great for autocomplete, but it doesn't know your organisation's code. It can't tell you how your team handles authentication or which service owns the payment flow.

The Solution: RAG + Copilot Extension

GitSage does three things:

Indexes — Crawls your GitHub org (READMEs, source code, issues) and stores embeddings in PostgreSQL + pgvector
Retrieves — When you ask a question, finds the most relevant code chunks via similarity search
Answers — Feeds the retrieved context to an LLM and streams a grounded response

The magic is that it also implements the GitHub Copilot Extensions protocol, so it plugs directly into the Copilot Chat experience your team already uses.

Architecture

graph LR
    A[Developer] -->|"@gitsage"| B[GitHub Copilot]
    A -->|curl| C[REST API]

    B -->|SSE| D[Copilot Extension]
    C --> E[Chat API]

    D --> F[RAG Engine]
    E --> F

    F --> G[Similarity Search]
    F --> H[LLM]

    I[Indexer] --> J[Chunker]
    J --> K[Embeddings]
    K --> L[(pgvector)]
    G --> L

Tech Stack

I chose this stack deliberately:

Choice	Why
Java 21	Records, sealed classes, virtual threads — modern Java is nice
Micronaut 4	Faster startup than Spring Boot, compile-time DI, GraalVM-ready
LangChain4j	Java-native RAG framework, no Python dependency
pgvector	Vectors in PostgreSQL — one less service to manage
OpenAI	GPT-4o for chat, text-embedding-3-small for vectors

The Copilot Extension Protocol

This was the most interesting part. GitHub Copilot Extensions use an OpenAI-compatible SSE streaming format. Your endpoint receives a list of chat messages and streams back tokens:

@Controller("/copilot")
public class CopilotExtensionController {

    @Post
    @Produces("text/event-stream")
    public HttpResponse<Publisher<String>> handleCopilotRequest(
            HttpRequest<String> httpRequest,
            @Body String rawBody) {

        // 1. Verify GitHub's webhook signature
        var signature = httpRequest.getHeaders()
            .get("X-Hub-Signature-256");
        if (!signatureVerifier.verify(signature, rawBody)) {
            return HttpResponse.status(HttpStatus.UNAUTHORIZED);
        }

        // 2. Extract the user's question
        var request = objectMapper.readValue(rawBody, 
            CopilotRequest.class);
        var userMessage = extractLastUserMessage(request);

        // 3. Stream RAG-augmented response
        Sinks.Many<String> sink = Sinks.many().unicast()
            .onBackpressureBuffer();

        ragService.chatStream(userMessage, token -> {
            sink.tryEmitNext(formatSSE(
                StreamEvent.token(token)));
        }).whenComplete((v, error) -> {
            sink.tryEmitNext(formatSSE(StreamEvent.done()));
            sink.tryEmitNext("data: [DONE]\n\n");
            sink.tryEmitComplete();
        });

        return HttpResponse.ok(sink.asFlux());
    }
}

Each SSE event follows the OpenAI chat completion chunk format:

data: {"id":"chatcmpl-gitsage","object":"chat.completion.chunk","choices":[{"index":0,"delta":{"role":"assistant","content":"The auth"}}]}

Smart Chunking

Naive character-based splitting destroys context. GitSage's chunker is code-aware — it splits at natural boundaries:

Markdown headers (#, ##)
Class/method declarations (public class, function, def)
Blank lines (paragraph breaks)

With configurable overlap so context isn't lost between chunks.

One-Command Setup

This was a hard requirement. If people can't try it in 2 minutes, they won't star it:

git clone https://github.com/rameshreddy-adutla/gitsage.git
cd gitsage
cp .env.example .env  # add your tokens
docker compose -f docker/docker-compose.yml --env-file .env up -d

# Index your org
curl -X POST http://localhost:8080/api/index

# Ask a question
curl -X POST http://localhost:8080/api/chat \
  -H "Content-Type: application/json" \
  -d '{"question": "How does error handling work?"}'

Docker Compose spins up GitSage + PostgreSQL with pgvector. That's it.

What I Learned

1. pgvector is underrated. You don't need Pinecone, Weaviate, or Qdrant for most RAG use cases. pgvector with an HNSW index inside PostgreSQL is fast, simple, and one less service to manage.

2. LangChain4j is mature. The Java RAG ecosystem has caught up. Swappable providers, streaming support, batch embedding — it just works.

3. Copilot Extensions are a blue ocean. Very few open-source examples exist. If you build one that's actually useful, the GitHub community notices.

4. README quality is 80% of stars. Architecture diagrams (Mermaid renders natively on GitHub), badges, and a clear quick-start section matter more than perfect code.

Roadmap

GitSage is just getting started:

🦙 Ollama support — run fully local, no API keys needed
🌐 Web UI — browser-based chat interface
📊 Prometheus metrics — observability for indexing and queries
⚡ GraalVM native image — instant startup
💬 Slack/Teams integration — chat from team channels

Try It

rameshreddy-adutla / gitsage

🧙 A sage that knows your codebase — RAG-powered GitHub org knowledge bot with Copilot Extension support

🧙‍♂️ GitSage

A sage that knows your codebase

Index your entire GitHub org → Ask questions about your code → Get AI-powered answers with source citations

Quick Start • Features • Architecture • Copilot Extension • Configuration • Contributing

💡 What is GitSage?

GitSage is a self-hosted RAG (Retrieval-Augmented Generation) bot that indexes your GitHub organisation's repositories and lets you chat with your codebase. It understands your code, READMEs, issues, and development patterns — and cites its sources.

Works as a GitHub Copilot Extension — type @gitsage in Copilot Chat and ask anything about your org's code.

You:      @gitsage how does authentication work in our services?

GitSage:  Based on the codebase, authentication is handled by the `auth-service` 
          repository using JWT tokens...
          
          📁 auth-service/src/main/java/com/example/AuthController.java
          📁 auth-service/src/main/java/com/example/JwtTokenProvider.java
          
          The flow is: Login → Validate credentials → Issue JWT → Store in 
          HTTP-only cookie → Verify on subsequent requests via JwtAuthFilter...

✨ Features

Feature

…

View on GitHub

The repo has:

Full source code (Java 21, Micronaut 4)
Docker Compose one-command setup
Detailed architecture docs with Mermaid diagrams
Copilot Extension setup guide
Contributing guide for newcomers

If this solves a problem for your team, give it a ⭐ — it helps others discover it.

What questions do you have about building RAG systems or Copilot Extensions? Drop a comment below — happy to go deeper on any part of this.