DEV Community: Shashanth Tarigopula

BhoomiAI — Bringing AI Closer to Indian Agriculture

Shashanth Tarigopula — Mon, 12 Oct 2026 06:54:41 +0000

This is a submission for the Hacktoberfest Week 1 dev Challenge: Touch some grass

What I Built

BhoomiAI is an open-source AI agriculture project I'm building to help farmers make more informed decisions using weather information, crop insights, and agricultural data.

The idea came from wanting to build something practical with open-source AI that solves a real-world problem rather than creating another application that keeps people glued to their screens.

With BhoomiAI, I want to bring useful agricultural information together in one place.

The planned features include:

🌦️ Weather Insights: Rainfall probability and forecasts for upcoming days to help farmers plan their agricultural activities.
🌾 Crop Insights: Crop recommendations based on location, season, and planting information.
📍 Location-Based Information: An interactive map of India to help users explore agricultural conditions in different regions.
📈 Market Price Insights: Current seed prices and estimated crop-market prices to help farmers understand potential costs and returns.
🧪 Pesticide Recommendations: Relevant crop-protection guidance based on crop conditions and identified problems.
🤖 AI Agriculture Assistant: An assistant that can help users understand weather forecasts and agricultural information in simple language.

The goal is to make these insights accessible through a simple interface instead of forcing farmers to navigate multiple sources.

BhoomiAI is currently under development. I'm building the interface first and plan to integrate reliable data sources and open-source AI as development progresses.

Demo

The project is currently under development, and I haven't deployed a live demo yet.

I'll share the working demo once the core dashboard and integrations are ready.

Pictures:

Code

GitHub Repository: https://github.com/Tsaishashanth/BhoomiAI

I'm building BhoomiAI in the open and plan to improve it incrementally as features are implemented.

How I Built It

I started with the product idea and interface, focusing on making agricultural information easy to explore.

My current development approach includes:

Next.js and React for building the web application.
TypeScript for maintainable and type-safe code.
Tailwind CSS for styling the interface.
Interactive dashboard design to organize agricultural information into separate, easy-to-understand sections.
Open-source AI and public data sources as the intended foundation for future intelligence and recommendations.

I'm prioritizing a clean interface and a practical user experience before adding the data integrations.

The next development steps are to connect weather and agricultural data, implement the map interactions, and explore suitable open-source AI models that can support the agriculture assistant without depending on expensive proprietary APIs.

Why Does Open Innovation Matter?

Agriculture is a field where useful information should be accessible to everyone, including people who cannot afford expensive software or AI services.

Open innovation gives projects like BhoomiAI the opportunity to build on publicly available datasets, open-source models, and community-developed tools.

It also allows developers to inspect implementations, contribute improvements, and adapt solutions to local needs.

For a project focused on Indian agriculture, this is especially important because weather conditions, crops, languages, and farming practices vary across regions.

My goal is to keep BhoomiAI accessible, transparent, and open to contributions while exploring ways to deliver useful agricultural insights with minimal infrastructure costs.

My Agent Session

I'll add my agent session here if I capture and publish one during development.

BhoomiAI — Touch Some Grass

Shashanth Tarigopula — Sun, 11 Oct 2026 16:04:28 +0000

The idea is simple: use maps, weather data, crop insights, and an AI assistant to help people better understand agricultural conditions across India.

I'm starting with the UI and will gradually work on integrating open-source AI models and free data sources.

More updates coming soon!

veyra — Build for a Friend

Shashanth Tarigopula — Mon, 05 Oct 2026 06:28:47 +0000

This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend

What I Built

I built veyra, a tool that helps you understand long educational YouTube videos without watching every minute.

The idea came from a problem I noticed while talking with a friend. Long educational videos may or may not contain a lot of useful information, but finding the exact parts you need can take a lot of time.

So I decided to build something around that problem.

With veyra, you paste a YouTube video URL and it breaks the video into 5-minute sections.

Each section gives you:

A topic
A short summary
Important points
Concepts mentioned
Useful resources

The goal is not to replace the original video. Instead, veyra gives you a structured way to explore the whole video and quickly understand what each part is about.

I also left the testimonials section open for feedback from the friends I built this for. I haven't received their reviews yet, so I decided not to add fake testimonials and will add their actual thoughts once they have tried veyra.

Demo

Live demo:

https://aiveyra.vercel.app

You can paste a YouTube video URL into the dashboard and let veyra analyze it.

Code

GitHub repository:

https://github.com/Tsaishashanth/veyra

The project is open source, so anyone can check the code or contribute.

How I Built It

The main flow of veyra is:

YouTube URL → Transcript → 5-minute sections → Gemma → Results

I built the application using Next.js, TypeScript and Tailwind CSS.

For the AI part, I used Gemma to analyze each 5-minute section of the video.

The transcript is first divided into smaller sections. Each section is then sent to Gemma, which generates the topic, summary, key points, concepts and useful resources.

I also had to handle videos where the normal YouTube transcript method doesn't work. For those cases, I added a fallback transcript service.

The main reason for splitting the transcript into smaller sections was to keep the AI focused on a specific part of the video instead of sending one large transcript and asking it to understand everything at once.

Why Does Open Innovation Matter?

I wanted to use an open model for veyra because I wanted the AI part of the project to be something I could experiment with and improve.

Using Gemma gives me the freedom to change the prompts, experiment with the way each section is analyzed and explore how open models can be used in a real application.

Since veyra is also open source, developers can look at how the AI is being used and contribute improvements to the project.

For me, open innovation makes it easier to experiment, learn and build on top of the work that already exists.

Prize Categories

Best Use of Gemma

Gemma is used as the main AI model in veyra to analyze each section of a YouTube video and generate the structured results.

Hacktoberfest Build Series - #4

Shashanth Tarigopula — Mon, 05 Oct 2026 02:52:37 +0000

The core dashboard flow for veyra is now working.

A YouTube video can now go through the complete pipeline: YouTube URL → Transcript → 5-minute sections → Gemma analysis → Structured results

Hacktoberfest Build Series - #3

Shashanth Tarigopula — Sun, 04 Oct 2026 21:39:46 +0000

YouTube transcript fetching is now working in veyra.

Pasted a YouTube URL → fetched the full transcript successfully.
Next up: breaking the transcript into 5-minute sections.

Hacktoberfest Build Series#2

Shashanth Tarigopula — Sun, 04 Oct 2026 17:02:58 +0000

Finished building the landing page for veyra.

Next, I'll be working on the dashboard and the core video analysis flow. More updates coming soon.

Build for a Friend — The Idea

Shashanth Tarigopula — Sun, 04 Oct 2026 13:38:00 +0000

I’ve been thinking about what to build for this challenge, and randomly remembered my friend saying that he doesn’t watch long-form educational YouTube videos.

So I took that as the idea for this challenge.

VEYRA

You give veyra a long YouTube video, and it uses AI to understand the video and break it into smaller sections. Each section will have a short summary, important points, concepts discussed, and useful resources.

More updates coming soon.

Build for a Friend – My Hacktoberfest Start

Shashanth Tarigopula — Sat, 03 Oct 2026 15:23:04 +0000

Hello everyone, this is my first post, I’m excited to join Hacktoberfest.

I started building "Build for a Friend" challenge.
Will share updates soon, stay tuned.