DEV Community: Khyati Kshitija

Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine

Khyati Kshitija — Tue, 27 Oct 2026 15:15:30 +0000

What I Built

🕯️ Ghost Journal AI: The Hollow — Where AI Meets the Supernatural

What if a haunted diary could create its own mysteries, hide secrets in evidence, and answer your questions while trying not to reveal the truth?

I built Ghost Journal AI: The Hollow, an AI-powered supernatural mystery investigation game. Players enter a fictional scenario, and the AI creates a mystery with three suspects, evidence clues, hidden motives, and a chilling story.

Players can investigate the suspects, examine the evidence, question a haunted diary, and finally accuse the culprit to reveal the solution.

The application also saves completed investigations in a local journal, so each case becomes part of the player's history.

I interpreted "Ghost in the Machine" as a story in which an AI feels almost supernatural: a voice inside a machine that knows the secrets of a mystery and guides you through the darkness.

Demo

The application currently runs locally on my laptop using Streamlit and Ollama.

Demo: I am preparing a video demonstration of the application.

The demo will show mystery generation, suspect investigation, evidence review, interaction with the haunted diary, and the final accusation.

Code

GitHub repository: Ghost Journal AI: The Hollow

The project is built with Python and Streamlit, uses Ollama for local AI inference, and stores investigation history in SQLite.

Replace the example GitHub URL above with your actual public repository link before publishing.

How I Built It

I built the application using open-source tools and a locally running open-weight language model.

My technology stack includes:

Python — application logic and investigation flow.

Streamlit — the interactive web interface.

Ollama — running the AI model locally.

Llama 3.1 8B — generating mystery stories and powering the haunted diary chatbot.

SQLite — storing investigation history locally.

Requests — communicating with Ollama's local API.

When a player enters a scenario, the application asks the model to generate a structured mystery containing a title, opening scene, three suspects, evidence clues, a diary entry, a culprit, a solution, a twist, and an epilogue.

The player can then question the haunted diary. After the player submits an accusation, the application reveals the culprit and solution and saves the result in the investigation journal.

I tested the main workflow, including case generation, suspect and evidence display, diary responses, accusations, and journal history.

Why Does Open Innovation Matter?

Open innovation made it possible for me to experiment with an AI-powered storytelling experience without building the entire project around a paid, hosted AI API.

By using Ollama and an open-weight model locally, the core AI experience can run on my own machine. After the model has been downloaded, the application can generate stories without requiring a continuous connection to a remote AI service.

This approach also gives me flexibility. I can experiment with different models, adjust prompts, and improve the mystery-generation process without depending entirely on one provider's hosted API.

It also comes with trade-offs: local models require sufficient hardware resources, and AI-generated mysteries can still contain inconsistencies. Improving the logical connection between evidence and solutions is one of my next development goals.

For me, open innovation means being able to learn, experiment, and build something creative with tools that I can run and adapt myself.

My Agent Session

I developed and tested the application locally.

I do not have a DevRelay session link to share at this time.

The Spirit of It

The idea behind The Hollow is simple: what if the machine isn't just answering you, but telling you a story it doesn't want you to finish?

The haunted diary is the heart of the experience. It responds to questions, hints at hidden truths, and keeps the atmosphere alive while the player searches for the culprit.

I wanted to combine the creativity of AI storytelling with the feeling of investigating an old supernatural mystery, where every clue matters and every character might be hiding something.

The project is also an experiment in making local AI feel more personal, interactive, and a little unsettling.

Sometimes the scariest thing isn't a ghost in an old house. It's a voice inside a machine that knows more than it should. 🕯️

My project explores open-weight AI and local inference through an interactive storytelling application.

Built with curiosity, open-source tools, and a haunted diary that refuses to stay silent.

hacktoberfest #hf26challenge #devchallenge #opensource #AI #Python #Streamlit #Ollama

Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady

Khyati Kshitija — Sat, 24 Oct 2026 07:20:22 +0000

What I Built

I built 🐦 Bird Buddy AI, an open-weight AI-powered birdwatching companion that helps people explore the birds they encounter outdoors.

The idea is simple: take a bird photograph, upload it to the app, and let an AI model help describe what is visible. The app also includes a digital bird journal where users can record sightings, add dates and locations, write notes, track statistics, and download their records as a CSV file.

Instead of manually maintaining a notebook or spreadsheet, users can keep their birdwatching observations organized in one place.

My goal was to combine AI experimentation with a small, practical tool that encourages people to observe and learn about nature.

Demo

🐦 Try Bird Buddy AI: https://bird-ai-o.streamlit.app/

Upload a bird photograph, explore the AI's response, and record your sightings in the journal.

Code

💻 GitHub repository: https://github.com/khyatikshitija/bird-ai

The repository contains the Streamlit application and its supporting code.

How I Built It

I built Bird Buddy AI using Python and Streamlit, with an open-weight vision-language model from Hugging Face.

The project uses:

Python for the application logic.

Streamlit for the interactive web interface.

Hugging Face Transformers to load and run the vision-language model.

SmolVLM-500M-Instruct for image-and-text-based bird exploration.

PyTorch for model inference, using CUDA when available and CPU otherwise.

SQLite to store bird sightings locally.

Pandas to organize journal data and support CSV export.

I also added journal statistics so users can see their total sightings, different birds recorded, and name corrections.

Building this project taught me that creating an AI application involves more than calling a model. I had to work through model-loading issues, deployment problems, database handling, and the challenge of making the model's answers more cautious when it cannot confidently identify a bird.

Why Does Open Innovation Matter?

Open innovation matters because it gives developers the opportunity to experiment, learn, and build useful tools without depending entirely on closed AI APIs.

Using an open-weight model allowed me to explore vision-language inference, integrate AI into my own application, and experiment with running the model locally. It also helped me learn about the trade-offs between model size, speed, available hardware, and identification quality.

Bird Buddy AI is a small project, but it demonstrates how accessible AI tooling can help people turn an idea into a working application.

It also reminded me that AI-generated identifications are not always correct. A responsible birdwatching companion should communicate uncertainty rather than present every prediction as a fact.

My Agent Session

I developed the project iteratively, using AI assistance to troubleshoot errors, improve the application, and work through deployment challenges.

Optional: Add your DevRelay agent-session link here if you have one.

The Lazy Payoff

Bird Buddy AI brings bird identification experiments and sighting records together in one place.

Instead of keeping separate notes and spreadsheets, users can record observations, review their statistics, and export their journal when needed.

The app is still an experiment, and its AI identification results need verification. My longer-term goal is to make it a more reliable and useful companion for curious birdwatchers.

Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle

Khyati Kshitija — Wed, 14 Oct 2026 19:57:48 +0000

🐦 Bird Buddy AI: Turning Open-Weight AI into a Birdwatching Companion

What I Built

I built Bird Buddy AI, a small project that brings AI and outdoor exploration together.

The idea was simple: what if someone could take a picture of a bird they spotted outside and use an AI-powered tool to start learning about it?

Bird Buddy AI lets users upload a bird photograph and receive an AI-generated identification suggestion with a description of visible features. It also includes a daily nature challenge to encourage people to spend time outdoors.

I wanted to build something that wasn't just another chatbot. I wanted to connect a digital tool with an activity that happens in the real world.

For now, it's a prototype for beginner birdwatchers, students, and nature enthusiasts. In the future, I could explore a freemium version for birdwatchers who want richer species guides, regional information, and more advanced identification tools.

Demo

🌿 Try the live application: https://bird-ai-o.streamlit.app/

Code

💻 Explore the source code: https://github.com/khyatikshitija/bird-ai

How I Built It

I built Bird Buddy AI with Python and Streamlit. For image analysis, I used SmolVLM-500M-Instruct, an open-weight vision-language model from Hugging Face, through PyTorch and Transformers.

The application accepts an uploaded image, processes it, and asks the model to generate a response about the bird. I also worked on the interface and a daily nature challenge to make the experience approachable.

One of my biggest learning moments came during testing. The model sometimes gave a confident-sounding answer that was actually incorrect. That showed me that generating a convincing explanation isn't the same as correctly identifying a species.

I'm treating bird identification as a suggestion rather than a scientific certainty, and improving species-level accuracy is one of my next goals.

The local version runs inference on my laptop after downloading the model. This lets me experiment without sending photographs to a separate inference API through the local application. The hosted demo uses Streamlit Cloud.

Why Does Open Innovation Matter?

As someone learning by building, access to open-weight AI made this project possible for me to experiment with directly.

I could run a model, inspect how it behaved on real photographs, build an interface around it, and learn from its mistakes without depending on a paid hosted inference API for local testing.

Open innovation also means the project can be examined, modified, and improved by other developers. A future contributor could help improve the identification process, test different models, or add support for more bird species.

For me, the value isn't only that an AI model can identify an image. It's that open technology gives more people the opportunity to learn, experiment, and build something useful.

Prize Categories

“I'm submitting to the overall challenge.”

devchallenge #hf26challenge

Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass

Khyati Kshitija — Sat, 10 Oct 2026 18:22:54 +0000

🐦 Bird Buddy AI: Explore Nature with Open-Weight AI

What I Built

Bird Buddy AI is an AI-powered birdwatching companion designed to encourage people to step outside, observe wildlife, and learn more about the birds around them.

Users can upload a bird photograph and receive an AI-generated identification suggestion along with a description of visible colours, shapes, and markings. The app also includes a daily nature challenge that encourages users to spend 10 minutes outdoors observing birds without disturbing them.

I built this project to combine AI experimentation with real-world nature exploration. Instead of using AI only for screen-based activities, Bird Buddy AI encourages people to use technology as a starting point for discovering the natural world.

Demo

🐦 Live app: https://bird-ai-o.streamlit.app/

Try uploading a bird photograph and see what the AI suggests!

Code

💻 GitHub repository: https://github.com/khyatikshitija/bird-ai

The project is built with Python and Streamlit, with an open-weight vision-language model running locally within the app's Python environment.

How I Built It

I built Bird Buddy AI using Python, Streamlit, PyTorch, Hugging Face Transformers, and Pillow.

The core AI model is SmolVLM-500M-Instruct, an open-weight vision-language model that can process images and generate text responses.

The workflow is simple:

The user uploads a bird photograph.

The app prepares the image for the model.

The vision-language model examines the image and generates a bird identification suggestion.

The app displays the response and reminds users that AI predictions can be incorrect.

A daily nature challenge encourages users to continue observing wildlife outdoors.

I used Streamlit to build the interface and Hugging Face Transformers and PyTorch to load and run the model. After the model downloads, inference runs within the app's Python environment rather than through a separate hosted AI inference API.

One important limitation is that a general-purpose vision-language model may misidentify similar bird species. The app is intended as a learning companion, not a definitive scientific identification tool.

Why Does Open Innovation Matter?

Open innovation made it possible for me to build and experiment with an image-understanding AI application using an openly available model and accessible development tools.

As a developer exploring AI, I could inspect the model's documentation, integrate it into my own Python application, and experiment with image inputs and prompts without depending on a proprietary hosted inference API.

Open-weight models also make experimentation more flexible: developers can explore different deployment environments, adapt their applications, and learn how AI systems work in practice.

For Bird Buddy AI, this openness helped me turn an idea into a working project that connects AI with outdoor learning. It also reminded me that AI should not only help us do more on our screens; it can encourage us to step away from them and engage with the world around us.

My Agent Session

This section is optional. I did not include an agent session link in this submission. If I save a DevRelay session, I can add it here.

Prize Categories

This project uses an open-weight AI model and open-source development tools. I will list the specific eligible partner categories once I verify them against the challenge rules.

Built with curiosity, open-weight AI, and a little encouragement to touch grass. 🌿
if doubt comment below...

devchallenge #hf26challenge

FIRST DAY...

Khyati Kshitija — Sat, 10 Oct 2026 15:37:11 +0000

Hi everyone! 👋😊

I'm excited to be part of the DEC Community! I'm here to learn new things, connect with like-minded people, share ideas, and grow my skills.

Recently, I've been exploring AI and building projects like my Deadline Tracker, an AI-powered app that helps students manage academic deadlines using AI and email digests. 📚✨

Looking forward to learning from all of you and being part of this amazing community!

Feel free to say hi! 🚀