DEV Community: Simran Jaiswal

Ofrenda: A Private Altar for the People We Miss, Powered by a Local Model

Simran Jaiswal — Tue, 27 Oct 2026 14:45:42 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine

What I Built

Ofrenda is a private altar builder for Día de los Muertos. You tell it about someone you miss: their name, who they were to you, their favorite foods and songs, and a story you carry. A language model running on your own computer turns those details into a short remembrance and a gentle four-line calaverita verse. The page then lays everything out as a candle-lit, three-tier ofrenda with flickering candles, a papel picado banner, and your foods and songs as offerings. You can save the finished altar as a single HTML file and keep it.

I read "Ghost in the Machine" as honoring the people who have gone, rather than haunting anyone. The "ghost" is a memory, and the machine's job is to help you put it into words. The model is told to use only the details you give it, never to invent facts, and never to speak as the person who died. It helps you say what you already feel, and it doesn't pretend to be them.

Demo

Live site: https://er-simran-jaiswal.github.io/Ofrenda/

The hosted page is static, so it shows the altar design. Generating the remembrance needs the local model, so for the full experience run it on your own machine (steps below). [Add a screenshot or short screen recording of a finished altar here.]

Code

[Paste your GitHub repo link here. DEV will embed it if you use {% embed https://github.com/er-simran-jaiswal/Ofrenda %}, assuming that is the repo URL.]

To run it locally:

ollama pull llama3.2
python3 ofrenda.py
# opens http://localhost:8484

To use a different model: OFRENDA_MODEL=gemma3 python3 ofrenda.py

How I Built It

Model: an open-weight model served locally by Ollama. It defaults to llama3.2, and you can swap in any model Ollama can run.
Backend: one Python file using only the standard library (http.server and urllib). It receives the form, builds a prompt, and calls Ollama's /api/chat endpoint on localhost. There are no pip installs, API keys, or accounts.
Prompting: the system prompt tells the model to be warm and specific, use only the details provided, and never speak as the deceased. The model returns the remembrance and the verse separated by ---, and the app splits them.
Frontend: a single HTML file with plain CSS and JavaScript, using no frameworks and no external fonts, so it still looks right offline. The candles use a CSS animation that respects prefers-reduced-motion.
Export: "Save altar as a file" bundles the altar and its styles into one standalone HTML file that doesn't depend on the app or the model.

Why Does Open Innovation Matter?

Privacy: grief and family stories are about as personal as data gets. With a local model, nothing leaves the machine, so there is no server log, no account, and no company holding someone's memories of their grandmother.
Offline: once the model is downloaded, Ofrenda works with the Wi-Fi off, like candles in a power cut.
Longevity: a remembrance shouldn't disappear because an API changed its pricing or shut down. The tool depends on no hosted service, and the saved altar is a plain HTML file that will open for years.
Control: I can swap the model, for example to one that writes better in Spanish, or edit the system prompt to change its tone. Neither is possible with a closed API.

My Agent Session

[Optional: embed your DevRelay session with the agent_session tag, or link to it. Delete this section if you don't have one.]

The Spirit of It

[Optional, and this is the section I'd most encourage you to fill in. If you try Ofrenda with a real memory (a recipe, a song, a story about someone you've lost), describe the moment it landed in your own words, even if it's just a few sentences. Remove this section if you'd rather not share.]

Prize Categories

[List the partner categories you're entering, or delete this section. I don't know which ones apply, so I haven't guessed.]

SlothFlow — Slow Down, Get Things Done

Simran Jaiswal — Wed, 21 Oct 2026 09:01:47 +0000

SlothFlow: An Open-Source AI-Powered Task Planner for Effortless Productivity

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady.

What I Built

Meet SlothFlow — a simple, AI-powered task planner inspired by one of nature's most relaxed creatures: the sloth! 🌿

We all have days when our to-do lists feel overwhelming. Between college assignments, coding projects, deadlines, and everyday responsibilities, even deciding where to start can take too much energy.

SlothFlow aims to solve this problem.

Users can enter their messy thoughts, tasks, and reminders into one text box. SlothFlow uses an open-weight AI model to turn that brain dump into a short, manageable action plan.

Key Features

AI-powered planning: Turns a messy to-do list into smaller, clearer steps.
Privacy-focused: Processes your text in the browser without sending it to a SlothFlow backend.
Open-weight AI: Uses FLAN-T5 Small through Transformers.js.
Simple task tracking: Check off completed tasks and copy your plan.
No paid AI API: No paid inference API or usage charges are required.

The goal is simple: spend less energy organizing your tasks and more energy actually doing them.

Demo

Live Website: [https://er-simran-jaiswal.github.io/SlothFlow-website-files/]

Project Poster: See the attached SlothFlow poster.

The website is designed with a calming sage-green and cream theme to make planning feel less stressful and more approachable.

The live demo will be available after deployment.

Code

GitHub Repository: [ https://er-simran-jaiswal.github.io/SlothFlow-website-files]

The project is built with HTML, CSS, and JavaScript, with Transformers.js providing access to the open-weight AI model.

The repository includes the website files, instructions for running the project locally, and an MIT license.

How I Built It

I built SlothFlow using:

HTML for the website structure.
CSS for the responsive layout and calming visual design.
JavaScript for task input, model inference, checklist generation, and interactions.
Transformers.js to run an AI model directly in the browser.
FLAN-T5 Small as the open-weight language model.

When a user enters a brain dump, the application sends the text to the model running in their browser and asks it to produce a short action plan.

The application also includes a basic checklist fallback if the AI model cannot load.

The first model download requires an internet connection. Afterward, browser caching may help avoid downloading the files again, although fully offline operation is not guaranteed.

Why Does Open Innovation Matter?

For SlothFlow, open innovation is about giving users more control over their tools, their data, and their costs.

1. More privacy

The application processes user input in the browser rather than sending it to a SlothFlow backend or a paid AI inference API. This gives users greater control over their personal notes.

2. Lower running costs

There are no paid AI API calls. This makes the project easier to experiment with and potentially maintain without usage-based inference costs.

3. Freedom to experiment

Using an open-weight model gives developers more flexibility to modify prompts, experiment with different models, and adapt the application to their needs.

4. Learning through open technology

As a beginner exploring AI and machine learning, I wanted to build something practical while learning how an AI model can be integrated into a real application.

Open-source tools made that exploration more accessible.

For this project, open innovation is not just about making the source code available. It is also about making AI-powered tools more approachable, adaptable, and affordable.

The Lazy Payoff 🦥

Imagine having ten tasks in your head and no idea where to begin.

Instead of spending extra time organizing everything manually, you can enter your tasks in one place and let SlothFlow help you decide on a manageable starting point.

The intended benefit is less planning friction, fewer overwhelming lists, and a clearer next step.

I have built the first version of the application, and the next step is to test the AI-generated plans with real task lists and improve the results based on what works.

Prize Categories

Open-Source AI
Open-Weight Models
Local / Browser-Based AI

Please confirm the official challenge's eligible prize categories before submitting.

Sometimes productivity isn't about doing more. It's about making the next step easier.

Slow down. Make a plan. Keep moving. 🌿

Hacktoberfest #OpenSource #AI #MachineLearning #TransformersJS #FLANT5 #WebDevelopment #BuildInPublic

HustleKit: Build AI-Powered Product Listings Locally with Open-Source AI

Simran Jaiswal — Thu, 15 Oct 2026 09:21:28 +0000

HustleKit: A Local-First AI Listing Writer for Small Businesses

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

What I Built

Meet HustleKit — Local Listing Studio, a small-business sidekick that helps independent sellers turn product details into shop-ready listings.

Writing product descriptions, catchy titles, and searchable tags takes time, especially for people managing a small online store on their own. Not everyone wants to pay for an expensive AI subscription just to write product listings.

HustleKit makes this process simpler. Sellers can enter their product name, category, selling platform, product details, brand voice, and ideal customer to generate:

Search-friendly product titles
Customer-focused product descriptions
Relevant search tags
Easy-to-copy results for their shop

The goal is to help small sellers spend less time writing listings and more time building their businesses.

Demo

Try HustleKit here:

https://er-simran-jaiswal.github.io/hustlekit-2/

The website includes a demo mode so you can explore the interface immediately. For actual open-weight model inference, connect a local Ollama model using the setup instructions.

Code

GitHub Repository:

https://github.com/Er-Simran-Jaiswal/hustlekit-2

The project is built using HTML, CSS, and vanilla JavaScript, with an optional connection to Ollama for local AI inference.

Feel free to explore the code, try it locally, and suggest improvements!

How I Built It

I built HustleKit using three core web technologies:

HTML5 for the structure and input forms.
CSS3 for the responsive layout and visual design.
JavaScript for handling user inputs, generating and displaying listings, copy-to-clipboard controls, and connecting to a local AI model.

For the AI component, HustleKit can connect to Ollama, which lets users run an open-weight model such as Llama 3.2 on their own computer.

To try local inference, install Ollama and run:

ollama pull llama3.2
ollama run llama3.2

Then serve the project locally, open HustleKit, and connect the local model through the app's connection panel. Depending on your setup, you may need to configure Ollama's browser-origin permissions.

HustleKit also includes a clearly labelled demo mode that uses templates and keyword extraction. This lets people try the interface without setting up a model, while keeping the distinction between demo output and actual AI-generated output transparent.

Why Does Open Innovation Matter?

For me, open innovation is about giving developers and users more control over the tools they build and use.

HustleKit explores how open-weight AI can make useful writing assistance more accessible to independent sellers.

Here's why the open approach matters:

Privacy: When connected to a local model, product details are processed on the user's computer instead of being sent to a hosted AI API.
Cost control: Users can avoid per-request charges from a hosted AI provider, although running a local model still requires hardware and electricity.
Flexibility: Users can experiment with different models supported by their local Ollama setup.
Less vendor lock-in: The application isn't tied to one proprietary AI API or its pricing.
Accessibility: The demo mode makes it easy to explore the project before configuring local AI.

A closed AI service may perform better for some tasks, but I wanted to explore an alternative that prioritizes privacy, flexibility, and user control.

This project helped me see how open-source tools can be used to build practical products for everyday problems—not just experimental AI demos.

What's Next?

HustleKit is a starting point, and there is plenty of room to improve it. Some features I'd like to explore next include:

Platform-specific listing limits for Etsy, eBay, and Shopify.
Saving reusable brand voice preferences locally.
Exporting completed listings.
Gathering feedback from real sellers to understand which features would be most useful.

My next goal is to get the tool in front of real users and learn how it could become genuinely valuable to their businesses.

Final Thoughts

Building HustleKit has been an opportunity to combine frontend development with local, open-weight AI and explore a practical side-hustle idea.

I hope this project encourages more developers to experiment with open tools and build technology that gives people greater choice and control.

I'd love to hear your feedback and ideas for improving HustleKit!

devchallenge #hf26challenge #opensource #AI #JavaScript #Ollama #WebDev

Trail Ears: Identify Bird Calls in Your Browser, Even With No Signal

Simran Jaiswal — Sat, 10 Oct 2026 14:42:24 +0000

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

What I Built

Trail Ears is a website that identifies bird calls. You record or upload a short clip, and it shows the top 3 likely species with confidence scores. You can log each sighting with a note and export your trail log as a CSV.

It is built to make the screen the shortest part of a walk: It keeps screen time short: record a clip of up to 12 seconds, put the phone away, and check the result and log it when you’re ready. It works for hikers, birders, run clubs, or anyone who has stood outside wondering what that sound is.

Demo

Live site: https://Er-Simran-Jaiswal.github.io/birdnet-web/trail-ears.html

Code

Repo: https://github.com/Er-Simran-Jaiswal/birdnet-web

The new pieces are trail-ears.html (the interface, location filter, and trail log) and sw.js (the offline cache).

How I Built It

Model: BirdNET, an open-source bird sound identification model from the Cornell Lab of Ornithology and Chemnitz University of Technology.
Inference: it runs in the browser with TensorFlow.js, using the open-source browser port georg95/birdnet-web, which I forked.
My part: a new front end with recording and upload, top-3 results, a minimum-confidence slider, an optional location filter that narrows species to the area and time of year, and a local trail log with CSV export.
Offline: a service worker caches the model and page after the first load, so it can keep working with no signal.

Why Does Open Innovation Matter?

No signal needed: a closed API stops working the moment you lose coverage, which on a trail is almost always. With open weights the model lives in the browser's cache.
Privacy: audio and location are processed on the device, so there is no server to send them to.
Cost: no API key, no account, no usage bill.
Hackable: because the model and the browser port are open, I could build a new interface and location filtering on top without asking anyone for permission.

Credits: BirdNET (CC BY-NC-SA 4.0, non-commercial use) and the TensorFlow.js browser port by georg95.

Leaf Detective: Explore Nature, One Leaf at a Time 🍃

Simran Jaiswal — Fri, 09 Oct 2026 11:25:58 +0000

🍃 Leaf Detective: Explore Nature, One Leaf at a Time

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

What I Built

I built Leaf Detective, a nature-themed project designed to encourage people to explore the world beyond their screens. The idea is to make learning about leaves and plants more engaging and interactive.

Instead of spending all your time online, Leaf Detective encourages you to step outside, observe nature, and discover the beauty of the plants around you.

This project is for students, nature lovers, and anyone who wants to reconnect with the outdoors through technology.

Demo

🌐 Try the live website:
https://er-simran-jaiswal.github.io/leaf-detective/

Code

💻 Explore the source code on GitHub:
https://github.com/Er-Simran-Jaiswal/leaf-detective

The project is open source, and you can explore the code, learn from it, and contribute improvements.

How I Built It

I built Leaf Detective as a web project and hosted it using GitHub Pages. I used GitHub to manage and publish the source code and make the website accessible online.

My goal was to combine technology with the natural world and create something that fits the spirit of the Touch Grass challenge.

AI tools and models used: [Add the exact open-source AI model, framework, or agent you used, if applicable.]

Why Does Open Innovation Matter?

Open innovation makes it possible for developers to learn from one another, share ideas, and improve projects collaboratively.

By making projects open source, developers can contribute improvements, discover new approaches, and build better solutions together. It also helps beginners like me gain practical experience, learn from the community, and turn ideas into real projects.

Leaf Detective is a small step toward learning by building and sharing.

My Agent Session

[Optional: Add your DevRelay agent session link here if you have one. Otherwise, remove this section.]

Prize Categories

[Add the applicable partner prize categories based on the official challenge rules. Remove this section if you are not entering any partner categories.]

Thanks for checking out Leaf Detective! 🌿