DEV Community: Sagar Devde

🪶 DevRelief — Make Work Feel Lighter

Sagar Devde — Sat, 24 Oct 2026 12:15:44 +0000

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

What I Built

🪶 DevRelief — Make Work Feel Lighter

DevRelief is a local-first, AI-assisted productivity workbench designed to reduce repetitive work across the software delivery lifecycle.

Developers spend too much time summarizing Git activity, investigating CI failures, preparing release notes, converting rough requirements into user stories, writing test scenarios, documenting bugs, and chasing action items. Product owners, Scrum Masters, QA engineers, and engineering managers face similar repetitive tasks.

DevRelief brings these workflows into one workbench, helping teams turn raw inputs into structured, evidence-backed outputs with less manual effort.

It includes workflows designed for five roles:

Developers: Git activity digests, change summaries, release notes, and CI failure triage.
Product Owners: User stories, Gherkin acceptance criteria, and requirements ambiguity analysis.
Scrum Masters: Meeting action items and retrospective theme analysis.
QA Engineers: Test scenario generation and structured bug reports.
Engineering Managers: Delivery summaries and blocker matrices.

The goal is simple: spend less time formatting, summarizing, and repeating work—and more time solving meaningful engineering problems.

Demo

🚀 Live application: https://devrelief.ai.studio/

Code

💻 GitHub repository: https://github.com/DEVDESAGAR1/devrelief

DevRelief is being developed with an open-source-first approach, focusing on privacy, reproducibility, and practical developer productivity.

How I Built It

DevRelief uses a web-first architecture built around React, TypeScript, Vite, and Tailwind CSS, with an Express backend for server-side integrations.

The project follows a milestone-based roadmap, progressing from the application foundation and workflow engine to multimodal ingestion, role-specific workflows, AI integration, workflow history, security hardening, and release readiness.

AI and Local Inference

Ollama and compatible Gemma models are part of DevRelief's local-inference experience. The integration is designed to discover locally installed models and run supported AI tasks through a local Ollama service without automatically downloading models.

DevRelief also provides a deterministic mode that uses application logic without requiring an AI API. For users who explicitly consent to cloud processing, the application includes a server-side Gemini integration.

Key design principles include:

Local-first processing: Keep supported work local whenever possible.
Choice of AI mode: Deterministic processing, local Ollama inference, or consent-based Gemini processing.
Evidence-backed outputs: Preserve source references and distinguish facts, inferences, and unknowns.
Privacy by default: Scan inputs for secrets and avoid sending content to cloud services without explicit consent.
Human control: AI-generated command suggestions remain suggestions and are never automatically executed.

The project is being developed incrementally, with automated tests and production-build checks accompanying implementation. Comprehensive browser-based UI testing and live-provider verification remain important release-readiness tasks.

Why Does Open Innovation Matter?

Open innovation matters because developer productivity tools should not force every user to depend on a proprietary AI service.

By building around open-source tooling and local inference, DevRelief aims to give developers more control over their data, model choices, operating costs, and working environment.

A developer should be able to use deterministic workflows without an API key, experiment with locally installed Gemma models through Ollama, and opt into cloud inference only when it makes sense.

Open tooling also makes the project easier to inspect, extend, test, and improve through community contributions. The long-term vision is a workbench that developers can adapt to their own workflows rather than a closed assistant they must work around.

My Agent Session

The project was developed iteratively with a coding agent, using milestone-based implementation, explicit acceptance criteria, automated tests, and build verification.

The Lazy Payoff

DevRelief targets repetitive tasks that interrupt engineering work: preparing summaries, structuring requirements, extracting action items, triaging failures, and drafting test cases.

Instead of starting from a blank page, users can begin with structured outputs generated from their supplied evidence and then review and refine the result.

The expected payoff is less repetitive manual work and more consistent documentation across software delivery.

I am continuing to validate the application and its AI integrations before making quantified time-saving claims. I would rather measure the actual benefit than invent a percentage.

Prize Categories

Gemma

🏢 SocietyFix AI — The AI-Powered Operating System for Housing Societies

Sagar Devde — Sun, 18 Oct 2026 14:24:55 +0000

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

🏢 SocietyFix AI — The AI-Powered Operating System for Housing Societies

What I Built

SocietyFix AI is an AI-powered housing society management platform designed to simplify the daily operations of residential communities in India.

Most housing societies still depend on WhatsApp groups, Excel spreadsheets, paperwork, and manual coordination for managing maintenance payments, complaints, meetings, vendors, and residents.

SocietyFix AI aims to bring these disconnected processes into one intelligent operating platform.

🚀 Key Features

🏘️ Society Management: Manage multiple societies, buildings, units, residents, and committee members.
👥 Role-Based Access Control: Secure access for administrators, committee members, and residents.
🤖 AI-Powered Assistance: Support document understanding, drafting, summarization, and operational recommendations.
📋 Complaint Management: Register, assign, track, and resolve resident complaints.
🔧 Asset & Maintenance Management: Track society assets, preventive maintenance, and service requirements.
👷 Vendor Management: Manage vendors, quotations, approvals, and work orders.
💰 Maintenance Billing: Generate maintenance bills, record collections, issue receipts, and reconcile payments.
📊 Financial Operations: Support expense management, invoice processing, budgets, and financial oversight.
📅 Meetings & Governance: Manage agendas, meeting minutes, decisions, tasks, and notices.
📂 Document Management: Organize society documents, notices, and compliance records.
🔐 Super Admin Control Center: Inspect platform modules, review access permissions, diagnose integrations, and track implementation gaps.

Some capabilities are part of the broader product roadmap and may still be undergoing implementation or verification.

💡 What Problem Does It Solve?

Housing societies face several recurring challenges:

Lack of transparency in financial and operational activities.
Delayed complaint resolution and poor follow-up.
Manual maintenance billing and reconciliation.
Fragmented communication between residents and committees.
Difficulty maintaining records, documents, and compliance information.
Repetitive administrative work that consumes committee members' time.

SocietyFix AI is designed to reduce this operational burden through centralized workflows, automation, and responsible AI assistance.

💵 Who Would Pay for It?

Potential customers include:

Cooperative housing societies
Residential apartment associations
Property management companies
Professional society management agencies
Residential communities managing multiple buildings

The proposed business model is subscription-based SaaS.

Potential pricing models include:

Per society subscription
Per residential unit subscription
Tiered plans based on features
Premium AI-assisted management capabilities

Pricing, customer demand, and willingness to pay are still to be validated.

The goal is simple: make professional society management accessible, efficient, and transparent.

Demo

🌐 Live Application: https://societyai.ai.studio/

Code

💻 GitHub Repository: https://github.com/DEVDESAGAR1/societyal

SocietyFix AI is built with a modular architecture designed for multi-tenant housing society operations.

🛠️ Technology Stack

Layer	Technology
Frontend	React, TypeScript, Vite
Styling	Tailwind CSS
Backend	Node.js, Express, TypeScript
Database	PostgreSQL 16
ORM	Drizzle ORM
Authentication	Firebase Authentication
File Storage	Google Cloud Storage
Deployment	Google Cloud Run
Infrastructure	Terraform
Background Processing	Google Cloud Tasks / Scheduler
CI/CD	Google Cloud Build

🏗️ Architecture Principles

Multi-tenant society isolation
Backend-enforced authorization
Role-based access control
Secure document storage
Auditable financial operations
Integer minor-unit financial calculations
Modular service architecture
Human approval for consequential AI-assisted actions

How I Built It

I started by treating SocietyFix AI as a real SaaS product rather than a standalone AI chatbot.

The development process was divided into milestones, each addressing a specific operational challenge faced by housing societies.

Phase 1 — Platform Foundation

Established the application architecture, frontend, backend, authentication, database models, society management, and access-control foundations.

Phase 2 — Society Operations

Developed workflows for:

Meetings and minutes
Task management
Notifications
Resident onboarding
Complaint management
Asset tracking
Preventive maintenance
Vendor management
Quotations and work orders

Phase 3 — Financial Management

Implemented maintenance billing and collection workflows covering:

Maintenance charge plans
Monthly bill generation
Payment recording
Receipt generation
Payment reconciliation
Vendor payables
Financial approval controls
Multi-currency support

The M12 implementation report recorded 33 passing milestone tests and 224 passing regression tests. These are reported development results rather than an independent production-readiness certification.

Phase 4 — AI-Assisted Operations

The AI architecture is designed to assist society administrators with information-heavy tasks such as:

Summarizing operational information
Drafting meeting minutes and notices
Extracting information from documents
Suggesting next actions
Supporting administrative decision-making

AI-generated recommendations are intended to remain subject to human review where financial, governance, or security consequences are involved.

🧠 Open-Source AI Implementation

Model: Gemma

Phase 5 — Quality and Platform Oversight

I also introduced a dedicated Super Admin Control Center milestone.

Its purpose is to inspect the platform itself:

Identify implemented and missing functionality.
Review roles and permissions.
Diagnose service integrations.
Record evidence-based findings.
Support safe role simulation.
Help prioritize improvements before release.

This reflects an important lesson from building larger applications: a successful build is not the same as a verified working product.

Why Does Open Innovation Matter?

Open innovation is important because housing society management involves sensitive financial records, resident information, operational decisions, and governance documents.

A platform handling this information should be transparent, adaptable, and independently auditable.

Open-source technologies provide several advantages.

🔓 Transparency

Open implementations allow developers and contributors to inspect how software behaves and identify potential problems.

🔐 Privacy and Control

Open-weight AI models can support self-hosted inference, giving organizations additional deployment options and greater control over where data is processed.

🧩 Flexibility

A modular architecture makes it easier to replace individual components, integrate different models, and adapt the platform to different housing societies.

🤝 Collaboration

Community contributions can help improve accessibility, security, localization, documentation, and society-specific workflows.

🚀 Reduced Vendor Lock-In

Open technologies make it easier to evolve the platform without depending entirely on one proprietary service.

For SocietyFix AI, open innovation is about building a practical product that can remain adaptable and trustworthy as it grows.

The Business Opportunity

SocietyFix AI addresses an everyday operational problem with the potential for recurring revenue.

Instead of selling AI as a novelty, the product aims to provide measurable value through:

Reduced administrative effort
Faster complaint handling
More organized financial records
Better operational visibility
Improved resident communication
Easier access to society information

The long-term vision is to build a scalable housing society management SaaS platform that can serve individual societies and professional property management organizations.

What's Next?

The development roadmap includes:

Completing and validating remaining operational modules
Expanding AI-assisted workflows
Preparing realistic demo data
Running end-to-end browser tests
Conducting a comprehensive UI/UX redesign
Strengthening security and reliability
Validating the product with real housing societies
Testing subscription pricing and customer willingness to pay

Final Thoughts

I built SocietyFix AI around a simple question:

What if managing a housing society could be as organized and transparent as managing a modern business?

Housing society committees spend significant time handling repetitive tasks, tracking payments, coordinating vendors, and responding to residents.

SocietyFix AI aims to give them one place to manage those responsibilities, with AI helping where it can genuinely save time.

This project is my attempt to turn an everyday problem into a practical, sustainable software product.

Built for Hacktoberfest 2026 — Open-Source AI Challenge, Week 2: Side Hustle.

Hacktoberfest #OpenSource #AI #SideHustle

🌱 GardenMitra Your Garden. Your Mitra.

Sagar Devde — Sun, 11 Oct 2026 19:23:03 +0000

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

🌱 GardenMitra — Your Green Friend

Grow something real. Touch grass.

What I Built

GardenMitra is an open-source garden companion that uses Gemma AI running locally through Ollama to help people understand their plants, plan growing spaces, and take practical gardening actions.

We spend hours searching for answers on our screens, even when what we really want is to get outside and do something meaningful.

Gardening can be intimidating for beginners. Which plants suit a balcony? Why might a plant be struggling? How should a growing space be organized? GardenMitra aims to make those decisions easier and turn gardening into a habit.

Whether you have a few pots on an apartment balcony, a terrace garden, or an indoor plant collection, GardenMitra is designed to make plant care more approachable.

🌿 What GardenMitra offers

AI gardening chat: Ask gardening questions and get assistance from Gemma running locally through Ollama.
Plant image analysis: Submit a plant or garden image for AI-assisted visual analysis.
Garden Doctor: Get structured guidance for common plant-care concerns.
Garden planning: Organize growing spaces and garden zones.
Sunlight intelligence: Use available observations and calculations to reason about growing conditions.
Care tasks: Break plant care into manageable actions.
Shopping list: Keep track of supplies needed for upcoming gardening activities.

GardenMitra combines AI assistance with deterministic gardening workflows. Not every recommendation or calculation is generated by an AI model.

The goal is to help users move from uncertainty to action: understand what needs attention, put the phone away, and spend time caring for their plants.

🌎 Why Touch Grass?

Technology should help us experience more of the world, not keep us looking at screens all day.

GardenMitra encourages a simple routine:

Check your garden and identify something that needs attention.
Ask for guidance when you need it.
Choose one manageable gardening task.
Step away from the screen and do it.
Return when you need to plan or record progress.

The best gardening session is not the one spent chatting with AI. It is the one that helps you grow something real.

Demo

Live application: https://gardenmitra.ai.studio/
GitHub repository: https://github.com/DEVDESAGAR1/gardenmitra
GitHub code scanning: https://github.com/DEVDESAGAR1/gardenmitra/security/code-scanning

Code

Repository: GardenMitra on GitHub

GardenMitra is being developed as a modular gardening application with a focus on practical workflows, understandable logic, and extensibility.

The repository is the primary reference for the implementation and setup instructions needed to run the application and reproduce the AI workflows.

How I Built It

I wanted to build something more useful than a generic gardening chatbot: an application that combines conversational assistance, visual input, and practical garden-management workflows.

🧠 Local AI with Gemma and Ollama

GardenMitra uses Gemma through Ollama for local AI inference.

The two AI workflows are:

Gardening chat: The user asks a gardening question and receives a response from the locally running model.
Image analysis: The user supplies a plant or garden image for AI-assisted analysis.

Running inference locally allows the AI to operate without sending every prompt to a hosted model provider, provided the application uses the local Ollama endpoint and no other component uploads that information.

The exact Gemma model version matters, particularly for image analysis. The repository should identify the model tag and document its supported capabilities.

🏗️ Engineering approach

The project combines AI-assisted workflows with conventional application logic:

Local Gemma inference through Ollama.
Text-based gardening assistance.
Image input for AI-assisted analysis.
Garden-space and zone organization.
Structured, rule-based care guidance.
Sunlight calculations where supported by available inputs.
Browser local storage for the initial experience.
Automated checks and production-build verification during development.

Weather data is simulated by the mock weather provider; GardenMitra does not claim that this provider supplies live weather information.

AI-generated plant guidance and image analysis should be treated as assistance, not a guaranteed diagnosis. Results can be incomplete or incorrect, so users should verify important plant-care decisions.

Reproducibility

AI runtime: Ollama.
Model family: Gemma.

These details help other developers reproduce the setup and understand which model capabilities GardenMitra uses.

Why Does Open Innovation Matter?

Gardening knowledge should be accessible, adaptable, and useful to people with different growing spaces, budgets, climates, and levels of experience.

Using open-weight AI locally gives GardenMitra an alternative to making every interaction dependent on a proprietary cloud API.

Privacy and control

With a local Ollama setup, prompts and images can be processed on the user's own machine instead of being sent to a hosted model provider. This depends on the actual deployment: other application services must also avoid uploading that information if the entire workflow is to remain local.

Accessible experimentation

An open-weight model gives developers the opportunity to experiment with different configurations and adapt the application to their needs.

Local inference can also avoid per-request charges from a hosted inference API, although it still requires suitable hardware and consumes computing resources.

Community-driven gardening

A balcony gardener in Mumbai may need different guidance from someone growing plants in a colder climate. An open-source application gives developers and gardening enthusiasts room to improve plant knowledge, refine workflows, and contribute region-specific improvements.

Open innovation is about more than choosing a model. It is about making the technology behind a useful tool easier to experiment with, adapt, and improve.

My principle is simple: technology should help people care for the real world, not replace their experience of it.

Prize Categories

Hacktoberfest Open-Source AI Challenge Week 1 — Touch Grass
Best Use of Gemma — intended category, subject to the official category rules and verification of the working implementation.

Final Note

GardenMitra brings together local AI, practical gardening workflows, and the belief that technology should encourage real-world action.

I built it to make gardening less intimidating and encourage people to spend more time growing, observing, and caring for plants.

Built by Sagar Devde.

🌱 GardenMitra — Grow something real. Touch grass.

devchallenge #hf26challenge #hacktoberfest #opensource #Gemma #Ollama #gardening #AI

ApnaGhar 🏠: Designing a Home That Feels Like Yours

Sagar Devde — Sun, 04 Oct 2026 18:12:22 +0000

What if planning your home didn't require expensive design software, specialised knowledge, or uploading every room photo to a third-party AI service?

That question inspired me to build ApnaGhar (अपना घर) — a home-design application that brings room planning, coordinated interior styling, 2D layouts, 3D visualisation, and AI-assisted room understanding into one experience.

🏠 Live Demo: https://apna-ghar.ai.studio/

💻 Source Code: https://github.com/DEVDESAGAR1/apnaGhar

🤝 The Idea: Building for a Friend

The idea behind ApnaGhar started when my friend wanted to design their house. Choosing wall colours, coordinating flooring, arranging furniture, and imagining the final result can be challenging without the right tools.

I wanted to make that process easier by creating a space where my friend could explore interior styles, experiment with colours, and visualise room layouts before making real-world decisions.

That thought became ApnaGhar — a project built around a simple idea: making home design more accessible through technology.

✨ What Is ApnaGhar?

ApnaGhar brings together three main experiences:

Visualise: Explore rooms through interactive 2D planning and 3D visualisation.
Coordinate: Experiment with architectural paint collections, colour combinations, interior styles, accent walls, trim, and flooring.
Understand: Explore AI-assisted room-photo analysis and furniture recommendations.

The goal is to connect creative design with practical room planning.

🛠️ What I've Built

🎨 A Whole-Home Design System

Curated interior styles, architectural paint collections, colour-family tagging, coordinated wall and flooring finishes, and room-level and whole-home design controls.

📷 AI-Assisted Room Understanding

Image inspection, indoor-room suitability validation, and furniture-review recommendations, with a tested local vision-processing workflow using Gemma and Ollama.

🧱 Interactive 2D and 3D Planning

A 2D room planner and Three.js-based 3D viewer that bring room layouts and material choices together.

🔒 Privacy-Conscious by Design

ApnaGhar explores local AI processing to help users maintain control over personal room photographs and reduce unnecessary dependence on external AI services.

🧰 Technology Stack

React, TypeScript, and Vite
Three.js for 3D visualisation
Interactive 2D room planning
Gemma and Ollama for local AI-assisted room understanding

Development checks included successful TypeScript checks and a production build with no reported errors.

🌱 Why Open Innovation Matters

Open-weight models give developers the freedom to experiment, adapt technology, and build useful applications without depending entirely on closed AI services.

With ApnaGhar, I wanted to explore how open-source AI could support everyday home-design decisions while keeping privacy and practical reliability in mind.

🚀 What's Next?

I want to continue improving room understanding, furniture recommendations, 2D planning, 3D visualisation, and the overall user experience based on real-world feedback.

❤️ Built for a Friend, Open to Everyone

ApnaGhar started with my friend's need to design their house and grew into an opportunity to explore how software engineering and open-source AI could make home planning easier.

For me, this project is about using technology to solve an everyday problem and helping people feel more confident about the place they call home.

🏠 Try ApnaGhar: https://apna-ghar.ai.studio/

💻 Explore the code: https://github.com/DEVDESAGAR1/apnaGhar

Thank you for checking out my project! I'd love to hear your feedback and ideas for improvement.

🏆 Challenge

Hacktoberfest DEV Weekend Challenge — Build for a Friend

Category: Best Use of Gemma

Hacktoberfest #WeekendChallenge #HF26Challenge #OpenSource #Gemma #HomeDesign #React #TypeScript #Ollama #BuildForAFriend