DEV Community: Sunkavalli Harshita

CaptionCraft AI ✨ — Your Little Social Media Content Studio

Sunkavalli Harshita — Sat, 17 Oct 2026 06:07:40 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle

What I Built

CaptionCraft AI is a simple social media content creation tool designed for small business owners, Instagram sellers, home bakers, boutique owners, and freelancers.

Creating engaging captions and finding relevant hashtags for every post can be time-consuming, especially for small businesses that cannot afford professional marketing services.

CaptionCraft AI aims to make this process easier by allowing users to select their business type, enter a product or post topic, choose a language, and select a preferred tone.

The goal is to help small businesses create attractive social media content faster and more affordably

Demo

Code

How I Built It

I built CaptionCraft AI using Python and Streamlit.

The application includes:

Business category selection

Product or post topic input

Language selection

Caption tone selection

A caption generation interface

An option to download generated captions

I initially explored Hugging Face Inference Providers for model access. However, I encountered inference credit limitations, which led me to explore local inference using Ollama and open-weight language models.

The local inference integration is still in progress.

Why Does Open Innovation Matter?

Open innovation makes AI experimentation more accessible to developers and small businesses.

Using open-weight models can offer more control over model selection, local deployment, and data privacy. A local inference approach can also reduce dependence on paid hosted inference APIs.

For a tool like CaptionCraft AI, these advantages matter because small businesses need affordable solutions, while users may prefer to keep their product information and draft marketing content on their own devices.

My goal is to build a practical content assistant around open AI technology rather than depend entirely on a closed, paid API.

My Agent Session

My First User

Prize Categories

Open-Source AI
Featured categories
Side Hustle

🌿 Scroll Less, Stroll More Building TrailMate AI — an AI-powered outdoor companion designed to turn screen time into real-world moments.

Sunkavalli Harshita — Tue, 06 Oct 2026 09:28:00 +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 companion built around a simple idea: technology should sometimes help us step away from technology. Instead of giving users another feed to scroll, it asks how much time they have, how they feel, and what kind of activity they prefer, then turns those choices into a personalized outdoor challenge—from a mindful walk to observing nature or simply slowing down and noticing the world around them. A Touch Grass Score adds a small sense of achievement by rewarding completed challenges based on time spent outdoors. Built for students, remote workers, and anyone caught in the cycle of endless screen time, TrailMate AI uses AI not as the destination, but as the nudge that gets people moving toward the real world. 🌿

Demo

https://trailmate-ai-frontend.onrender.com

Code

https://github.com/harshitasunkavalli/TrailMate-AI

How I Built It

TrailMate AI is built around the open-weight Qwen2.5-0.5B-Instruct model, accessed through Hugging Face Inference Providers with Featherless AI for lightweight, prompt-driven generation. A FastAPI backend receives the user's available time, mood, and preferred activity, turns those choices into a structured prompt, and asks the model to create a short, safe, personalized outdoor challenge. The experience is presented through a simple HTML, CSS, and JavaScript interface, while the Touch Grass Score is stored locally in the browser to keep the experience lightweight and account-free. The frontend and backend are deployed separately on Render, creating a simple architecture where AI becomes the bridge between a user's digital interaction and a real-world outdoor activity.

Why Does Open Innovation Matter?

Open innovation made it possible to build TrailMate AI around a model that I could choose, experiment with, and integrate into my own application rather than treating AI as a black-box service. Using an open-weight model gave me greater flexibility over the model layer and allowed the project to remain centered on a transparent, replaceable AI component instead of being tightly coupled to one proprietary ecosystem. More importantly, it changed how I think about AI: the goal was not to build another experience that keeps people online, but to use accessible AI technology to create a small interaction whose purpose is to end with the user stepping away from the screen.

My Agent Session

I used an AI-assisted development workflow while building TrailMate AI, particularly for iterating on the application structure, debugging integration issues, refining prompts, and preparing the project for deployment. The development process was iterative: test the idea, identify what failed, adjust the implementation, and test again until the complete frontend → backend → model workflow worked reliably.

Prize Categories

Which partner categories are you entering?
Featured categories, Partner categories

CareerMate AI — A Resume & Career Assistant Built for a Friend

Sunkavalli Harshita — Sat, 03 Oct 2026 11:44:30 +0000

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

What I Built

I built CareerMate AI, an AI-powered resume and career assistant designed for a friend who was struggling with one common problem:

“I have a resume, but how do I know whether it actually matches the job I’m applying for?”

Job applications can be overwhelming. You have to repeatedly compare your resume with job descriptions, identify missing skills, understand your ATS compatibility, and figure out what you should learn next.

So I wanted to build something that could make this process simpler.

CareerMate AI allows a user to:

📄 Upload their resume
🎯 Compare the resume with a job description
📊 Generate an ATS-style compatibility score
🔍 Extract relevant technical skills
❌ Identify missing skills from the job description
💼 Recommend suitable job roles
📚 Suggest skills/topics to learn
💡 Generate personalized resume feedback
📥 Generate a downloadable analysis report

Instead of manually checking everything for every job application, the goal is to give the user a quick, personalized career analysis in one place.

Demo

resume-analyzer-1-pe8f.onrender.com

Code

harshitasunkavalli / resume-analyzer

Working Mechanism: User Uploads the resume , then the system exctract the skills from the resume and checks with description provided meaning, resume skills are checked with description provided skills then it gives the ats score, matched score, job recommendations and ai provided feedback of the resume and a report to download.This platformalso contains hte plafroms like unstop, naukar to apply for job. deployment link : https://resume-analyzer-1-pe8f.onrender.com

Every useful for the buddies who are attending the placements session.

View on GitHub

How I Built It

The application is built using:

🐍 Python
🌐 Flask
📊 Pandas
🔎 Resume parsing & skill extraction
🧠 AI-powered analysis
📄 ReportLab for downloadable reports
🎨 HTML, CSS & JavaScript
🔧 Git & GitHub
🚀 Render for deployment

The application processes the uploaded resume, extracts the resume text and technical skills, and then compares those skills against the provided job description.

The analysis produces an ATS-style score, job-description match score, missing skills, career recommendations, and learning suggestions.

Resume PDF
↓
Resume Text Extraction
↓
Skill Extraction
↓
┌───────────────────────┐
│ │
ATS Analysis Job Description
│ │
└───────────┬───────────┘
↓
Match Analysis
↓
Missing Skills
↓
Career Recommendations
↓
Learning Suggestions
↓
PDF Report

Why Does Open Innovation Matter?

My Agent Session

Open innovation makes it possible for developers to experiment, build, modify, and share solutions without being limited to a single closed ecosystem.

For a project like CareerMate AI, this matters because career tools should be accessible, transparent, and customizable.

Using open technologies allows me to understand how each part of the system works, modify the analysis logic according to user needs, and continue improving the project based on real-world feedback.

More importantly, the project is inspired by a real problem faced by someone close to me rather than being built only as a theoretical application.

Prize Categories

AI / Open Innovation

devchallenge, #weekendchallenge, #hf26challenge.

From Hours to Seconds: How AI Is Changing the Way We Work ⚡

Sunkavalli Harshita — Sat, 03 Oct 2026 10:44:34 +0000

Have you ever worked on a task for hours and suddenly thought:
There has to be a faster way to do this.
think most developers have experienced this.

As developers, we spend a lot of time writing code, debugging errors, searching documentation, analyzing data, writing SQL queries, creating documentation, and doing many other repetitive tasks.

But today, AI is changing the way we approach this work.

Not by simply doing everything for us — but by helping us move from “How do I start?” to “How can I improve this?” much faster.

⏳ The Traditional Way

Imagine you are given a task:

“Analyze this dataset and find the important patterns.”

Without AI, you might:

Understand the dataset
Search for the right Python libraries
Write the code
Run it
Find errors
Debug the errors
Try different approaches
Create visualizations
Interpret the results

Depending on the task, this can take hours.

And sometimes, a large part of that time isn't spent solving the actual problem.

It's spent searching, debugging, and figuring out where to begin.

🤖 Enter AI

Now imagine explaining the same requirement to an AI assistant.

You could ask:

I have a CSV dataset.

Help me:

understand the columns
identify missing values
suggest useful visualizations
calculate important statistics
provide Python code for the analysis

Within seconds, you can have a starting point.

But here's the important part:

The AI-generated result isn't the final answer.

It is the starting point.

You still need to understand the code, verify the results, test it with your data, and make the necessary changes.

⚡ From Hours to Seconds

AI can be especially useful for repetitive development tasks.

For example:

🐛 Debugging

Instead of spending a long time searching through error messages, you can provide the error and relevant code:

Here is my Java code and the error I'm getting.

Explain:

Why this error occurs
Which line is causing it
How to fix it
How I can avoid this error in the future

The response can give you a direction almost instantly.

📝 Writing Documentation

Writing documentation from scratch can also take considerable time.

AI can help turn technical notes into a structured README:

Project
↓
Features
↓
Technologies Used
↓
Installation
↓
Usage
↓
Future Improvements

Instead of starting with a blank page, you start with a draft that you can edit.

💻 Understanding Code

Sometimes we don't need AI to write code.

We simply need someone — or something — to explain code we don't understand.

For example:

Explain this function line by line
and give me a simple real-world example.

This can make learning unfamiliar code much faster.

🧠 But AI Doesn't Replace Understanding

This is probably the most important lesson I've learned.

AI can generate code in seconds.

But understanding why that code works is still our responsibility.

AI can sometimes:

Generate incorrect code
Make assumptions about requirements
Provide outdated information
Miss edge cases
Produce code that looks correct but fails in real situations

So blindly copying AI-generated code isn't really development.

A better workflow is:

Problem
↓
Ask AI for possible approaches
↓
Understand the solution
↓
Write / modify the code
↓
Test it
↓
Debug
↓
Improve

The developer remains responsible for the final result.

🚀 AI Is Changing the Developer Workflow

I don't think the biggest advantage of AI is simply “writing code faster.”

The bigger advantage is reducing the amount of time we spend on repetitive work.

That gives us more time for things that require human thinking:

Problem solving.
Creativity.
Architecture.
Decision making.
Learning.
Building.

Instead of spending 2 hours figuring out how to start, you might get a useful starting point in a few seconds and spend those remaining hours improving the actual solution.