DEV Community: Ritik Kumar

I Built Udaan AI -A Multilingual AI Mentor for Indian Students Using Gemma 4

Ritik Kumar — Sun, 24 May 2026 17:22:12 +0000

This is a submission for the *Gemma 4 Challenge: Build with Gemma 4*

What I Built

Udaan AI is an AI-powered multilingual platform designed to help Indian students and underserved communities access career guidance, government schemes, scholarships, internships, and educational support through natural conversations.

The idea came from a simple problem:

Millions of students in India struggle to:

understand government schemes,
find scholarships,
choose career paths,
and access reliable guidance because of language barriers and lack of mentorship.

Most platforms are either:

too complicated,
English-heavy,
or difficult to navigate for beginners.

I wanted to build something that feels:

simple,
accessible,
conversational,
and genuinely helpful.

That became:

Udaan AI

An AI mentor that helps users:

explore careers,
discover opportunities,
understand schemes,
and learn skills using AI.

The platform supports:

English,
Hindi,
and Hinglish interactions.

Core Features

AI Mentor Chat

Users can ask:

“How do I become a web developer?”
“Which scholarships can I apply for?”
“Explain PMKVY scheme”
“Suggest a roadmap after BCA”

The AI responds conversationally and simplifies complex information.

Government Scheme Explorer

I created a searchable dashboard for:

scholarships,
startup schemes,
employment programs,
education initiatives,
rural development schemes.

The UI was designed to make government information easier to browse and understand.

Jobs & Internship Dashboard

Users can explore:

career sectors,
salary insights,
internships,
and recruitment pathways.

Personalized Dashboard

The platform includes:

saved schemes,
chat history,
uploaded documents,
and profile personalization.

Modern AI UI Experience

I focused heavily on creating:

a clean dark UI,
responsive layouts,
startup-style interfaces,
and intuitive navigation.

I wanted the product to feel modern and polished rather than just another AI wrapper.

Demo

Main AI Mentor Interface

Government Scheme Explorer

Jobs & Internship Dashboard

User Dashboard

Live Demo

[Add Deployment Link Here]

Code

GitHub Repository

How I Used Gemma 4

Gemma 4 powers the conversational intelligence behind Udaan AI.

I used Gemma 4 for:

multilingual conversations,
career guidance,
educational explanations,
government scheme simplification,
and contextual reasoning.

I chose the Gemma 4 31B Dense model because the project required:

stronger reasoning,
better long-context understanding,
and more natural conversational responses.

Since Udaan AI handles:

educational guidance,
scheme explanations,
and career planning, the larger context window and reasoning quality of the 31B Dense model made it the best fit.

Gemma 4 helped make the assistant feel:

more human,
more context-aware,
and more useful for real-world educational support.

Challenges I Faced

Some of the biggest challenges were:

structuring government scheme data,
designing intuitive UX flows,
balancing simplicity with functionality,
and making the AI responses feel genuinely helpful.

I also wanted to avoid building:

“just another chatbot.”

Instead, I focused on creating an AI product with real social impact.

What I Learned

While building Udaan AI, I learned:

AI product design,
prompt engineering,
frontend architecture,
UX thinking,
and the importance of accessibility in AI systems.

This project completely changed how I think about building AI products.

Future Improvements

I plan to add:

voice interaction,
more regional language support,
scholarship eligibility matching,
resume analysis,
AI interview preparation,
and WhatsApp integration.

Final Thoughts

AI becomes truly meaningful when it helps real people solve real problems.

Udaan AI is my attempt to make:

education,
career guidance,
and opportunities

more accessible for India’s next generation of learners.

Thanks for reading.

I Built Udaan AI -A Multilingual AI Mentor for Indian Students Using Gemma 4

Ritik Kumar — Sun, 24 May 2026 14:59:05 +0000

I Built Udaan AI — A Multilingual AI Mentor for Indian Students Using Gemma 4

This is a submission for the Gemma 4 Challenge: Build With Gemma 4

What Inspired Me

Millions of students in India struggle with:

understanding government schemes,
finding scholarships,
choosing career paths,
and accessing opportunities due to language barriers and lack of guidance.

Most platforms are:

difficult to navigate,
English-heavy,
or too overwhelming for beginners.

I wanted to build something that feels simple, accessible, and helpful — especially for students from underserved or rural backgrounds.

That idea became:

Udaan AI

An AI-powered multilingual mentor designed to help users:

explore careers,
discover scholarships,
understand government schemes,
and learn through natural conversations.

What Udaan AI Does

Udaan AI combines:

conversational AI,
educational assistance,
career guidance,
and government scheme discovery.

The platform supports:

English,
Hindi,
and Hinglish interactions.

Core Features

AI Career Mentor

Users can ask questions like:

“How do I become a web developer?”
“What skills should I learn after BCA?”
“Suggest a roadmap for freelancing.”

The AI generates:

learning roadmaps,
skill suggestions,
career paths,
and curated guidance.

Government Scheme Explorer

Users can browse and understand:

scholarships,
employment programs,
startup schemes,
education initiatives,
rural development opportunities.

The goal was to simplify access to information that often feels complicated.

Jobs & Internship Dashboard

I also added:

job categories,
salary insights,
internship guidance,
and career exploration features.

AI Chat Experience

The platform includes a conversational interface powered by Gemma 4.

Users can:

ask questions naturally,
upload documents,
and receive simplified explanations.

Why I Chose Gemma 4

Gemma 4 was the perfect choice because of:

strong reasoning capabilities,
multilingual support,
long-context understanding,
and flexibility for educational assistance.

I wanted the AI to feel conversational and helpful rather than robotic.

Gemma 4 made that experience much more natural.

Tech Stack

Frontend:

Next.js
Tailwind CSS

Backend:

Node.js

AI:

Gemma 4
OpenRouter API

Deployment:

Vercel

UI & Design

I focused heavily on creating:

a clean dark interface,
accessible navigation,
modern dashboard layouts,
and a startup-style user experience.

The goal was to make the platform feel welcoming and easy to use.

Screenshots

Main AI Mentor Interface

Government Schemes Dashboard

Jobs & Internship Explorer

User Dashboard

Challenges I Faced

Some of the biggest challenges were:

structuring scheme data,
designing intuitive UI flows,
creating meaningful AI prompts,
and balancing simplicity with functionality.

I also wanted to avoid making “just another chatbot” and instead create something purpose-driven.

What I Learned

While building Udaan AI, I learned:

AI product design,
prompt engineering,
frontend architecture,
UX thinking,
and how important accessibility is in AI systems.

This project changed how I think about building with AI.

Future Improvements

Some features I want to add next:

voice interaction,
regional language expansion,
scholarship eligibility matching,
resume analysis,
and WhatsApp integration.

Final Thoughts

AI becomes truly impactful when it helps real people solve real problems.

Udaan AI is my attempt to make education, career guidance, and opportunities more accessible for India’s next generation of learners.

Thanks for reading.

⚔️ Which Anime Coder Are You? Step Into the CodeSlayer Arena!

Ritik Kumar — Mon, 13 Oct 2025 14:30:37 +0000

Every coder has their own breathing style — but if you entered the CodeSlayer Arena, who would you become? 👾

From late-night bug battles to last-minute deployments, we’ve all got our inner anime warrior. So…
Pick your fighter:

🔥 Tanjiro – The Determined Innovator
Never quits — even when the code crashes for the 99th time.

⚡ Zenitsu – The Speed Coder
Writes 100 lines in 10 minutes… and prays it compiles. 😅

🎭 Nezuko – The Silent Creator
Calm, focused, and drops designs that leave everyone speechless.

💀 Inosuke – The Debug Slayer
Fights bugs with pure chaos and caffeine. Sleep is a myth. ☕

👇 So tell us — which one are you?
Tag your squad and assemble your CodeSlayer team!

🎯 Event Details

📍 Venue: National Institute of Technology (NIT) Delhi
📅 Date: 1–2 November 2025
🤝 In Association With: Unnat Bharat Abhiyan Cell, NIT Delhi

🛡️ Organized by DevSphere India

🚀 Register Now
👉 https://unstop.com/hackathons/code-slayer-2k25-national-institute-of-technology-nit-delhi-1560699?lb=lEixbRgS&utm_medium=Share&utm_source=shivakum42873&utm_campaign=Innovation_challenge

🏷️ Tags

CodeSlayer2K25 #Hackathon #AnimeVibes #DevSphereIndia #NITDelhi #CodingCommunity #Developers #TechEvents

🚀 Built My Developer Portfolio with Next.js + TypeScript + Tailwind CSS

Ritik Kumar — Thu, 28 Aug 2025 05:38:52 +0000

Hey everyone!
I’m excited to share that my developer portfolio website is finally live! This project is more than just a website — it’s my space to showcase who I am, the projects I’ve built, and my journey as a web developer.

🌟 Why I Built It

As a web developer, I wanted a single place where I could:
Showcase my projects and skills
Share my journey and experiences
Build a professional identity online
So, I decided to challenge myself and build my portfolio using modern web technologies.

🔧 Tech Stack

Here’s what I used to bring it to life:
Next.js (React Framework)
TypeScript (.tsx files for type safety)
Tailwind CSS (for sleek, responsive design)

📌 Features Inside

Home Page – clean and minimal introduction
About Me – my background and developer journey
Projects – highlights of my work
Skills – tools and technologies I’m confident with
Contact – ways to connect with me

🔗 Live Portfolio

👉 Check it out here: [https://spiky-portfolio-ojjp.vercel.app/
]

💡 What’s Next?

This is just the start. I’ll keep updating my portfolio as I learn and build more projects.
I’d love your feedback and suggestions to make it even better!

✨ Thanks for reading — and if you’re also working on your portfolio, I’d love to see it! Drop a link in the comments 🚀

CareerFinder: Find Your Career Path with Real-Time AI & RedisAI

Ritik Kumar — Sun, 10 Aug 2025 14:00:05 +0000

This is a submission for the Redis AI Challenge: Real-Time AI Innovators.

What I Built

CareerFinder is a web application that suggests career paths based on a user's skills, experience, and education.

The app uses a React frontend for an interactive form and a Node.js + Express backend to process inputs.

Predictions are powered by RedisAI with an ONNX-based model for real-time inference.

If the AI model is unavailable, the system gracefully falls back to mock predictions to maintain usability.

Key features:

Interactive career suggestion form
Real-time AI predictions via RedisAI
Mock fallback for offline model scenarios
Clean, responsive UI built with React

Demo

Live Demo: [coming soon...]

GitHub Repo: [https://github.com/Mr-spiky/CareerFinder

Screenshots:

How I Used Redis 8

CareerFinder uses Redis 8 with RedisAI as the real-time AI inference engine.

Model Storage: The ONNX model is stored in RedisAI for fast, in-memory access.
Real-Time Inference: When a user submits the form, the backend sends a tensor to RedisAI, runs the model with AI.MODELRUN, and retrieves predictions instantly.
Fallback Logic: If RedisAI or the model is unavailable, the backend switches to a mock prediction generator to ensure the app still returns results.

This setup ensures that career suggestions are delivered in milliseconds, providing a smooth, real-time experience.

“NexWork – A Smart Intranet for Modern Teams (Frontend Challenge Submission)”

Ritik Kumar — Mon, 28 Jul 2025 06:22:49 +0000

This is a submission for Frontend Challenge: Office Edition sponsored by Axero, Holistic Webdev: Office Space

🛠️ What I Built

I built NexWork, a smart intranet prototype designed for modern digital office spaces. The goal was to create a clean, responsive layout that fosters communication, collaboration, and engagement within teams — just like a real-world internal platform would.

With components like announcement cards, employee highlights, quick links, and event boards, NexWork reimagines traditional intranets as more interactive, aesthetically pleasing spaces.

🔗 Demo

Live Site: https://nexwork-project.vercel.app

GitHub Repo: github.com/Mr-spiky/nexwork-project

✨ Journey

This project gave me the chance to apply and improve my frontend development skills, especially using React, Tailwind CSS, and responsive design techniques. Here's what stood out:

🧩 Component-Based Design: I built reusable UI sections like info cards, team announcements, and navigation tiles using functional components in React.
🎨 Modern UI/UX: I focused on clean typography, soft shadows, and spacing using Tailwind utilities to reflect a relaxed but professional office vibe.
💡 Responsive Layout: The site adapts beautifully across devices to simulate a real company’s internal portal.

What I Learned:

Structuring layouts using flex and grid in Tailwind
Optimizing responsiveness and mobile-first design
Writing cleaner, scalable React components

🧑‍💻 Team Submission

This was a solo project created and coded by me.

⚠️ License

MIT License – free to use or remix with credit.

🙌 Final Thoughts

NexWork represents my vision of how intranets can be modern, useful, and even enjoyable. Thank you DEV and Axero for this opportunity to blend creativity and code with a meaningful purpose!