DEV Community: Akshat Bansal

TrailMate AI: Turn Screen Time into Green Time 🌿

Akshat Bansal — Sun, 11 Oct 2026 15:34:05 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass.

What I Built

TrailMate AI is an AI-powered outdoor adventure planner designed to help people spend less time scrolling and more time exploring the real world.

The idea is simple: users choose their available time, fitness level, and interests, and the application helps generate a personalized outdoor activity plan.

Whether it's a short nature walk, exploring a garden, or observing the surroundings in a local park, TrailMate AI aims to make getting outside easier and more enjoyable.

Demo

Live demo: Not available yet.

Demo video: Coming soon.

Code

GitHub Repository: https://github.com/akshatbnsl426/Trailmate-AI

How I Built It

I used Codex to help build TrailMate AI with HTML, CSS, and JavaScript for the frontend, Python and FastAPI for the backend, and Ollama for local inference with an open-weight language model.

The project includes automated tests, configuration examples, and setup documentation.

The test suite passed 14 tests in the development environment. Running the application with the local model and verifying the complete user experience are still part of my testing process.

Why Does Open Innovation Matter?

Open-weight AI gives developers the opportunity to experiment with models locally instead of depending entirely on proprietary AI APIs.

TrailMate AI is designed around local inference, allowing developers to experiment with prompts, change compatible models, and keep user inputs on their own machine when no external service is involved.

I also wanted to explore how AI can be used for something beyond keeping people on a screen. Instead, it can help people plan activities that encourage them to step outside, observe nature, and reconnect with the world around them.

My goal is to make AI a tool that encourages more real-world experiences, not more screen time.

My Agent Session

A DevRelay session link is not available yet.

Prize Categories

No partner-specific prize category claimed at this stage.

Study b : A Private Local AI Practice Partner for a College Friend

Akshat Bansal — Sat, 03 Oct 2026 20:45:08 +0000

StudyB: A Private Local AI Practice Partner for a College Friend 🎓

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

What I Built

I built StudyBuddy, a small AI-powered practice partner for a college friend preparing for technical interviews, vivas, and exams.

The problem was simple: sometimes you want to practice technical questions, but there isn't always someone available to practice with.

StudyBuddy provides three focused modes:

Viva Practice — practices questions one at a time like a viva examiner.
Interview Practice — focuses on technical interview topics such as Java, DSA, DBMS, and networking.
Explain a Topic — explains difficult concepts in simpler language with examples.

I wanted to build something small that solves one real problem rather than another generic chatbot.

Demo

The project is designed to run as a local Streamlit application.

A short demonstration of the application and its workflow is included with the project materials.

Code

GitHub repository:

https://github.com/akshat0042005/StudyBuddy

How I Built It

StudyBuddy is built with:

Python
Streamlit
Ollama
Qwen2.5 1.5B Instruct
Local HTTP inference

The architecture is intentionally simple:


text
User
  |
  v
Streamlit UI
  |
  v
Python Application
  |
  v
Ollama Local API
  |
  v
Qwen2.5 1.5B
  |
  v
Practice / Explanation / Feedback

hello everyone

Akshat Bansal — Sat, 03 Oct 2026 20:37:30 +0000

new to community