DEV Community: Saketha Rama

Ember: tend the things you're building, and watch them burn ❤️‍🔥

Saketha Rama — Sun, 12 Jul 2026 20:08:12 +0000

This is a submission for the Weekend Challenge: Passion Edition

Ember: tend the things you're building, and watch them burn

What I Built

Ember is a passion tracker for the things people want to keep alive: writing, running, music, learning, or a side project.

The idea came from a frustration with traditional habit trackers. A missed day can turn a useful tool into a guilt machine: the streak breaks, the counter goes back to zero, and the work that came before it feels invisible.

Ember uses a different metaphor. Every passion becomes a living flame. You give it a name, a color, and a reason for existing. When you show up, it grows through five stages:

Spark -> Flame -> Blaze -> Beacon -> Phoenix

The goal is not to be perfect. The goal is to make returning feel worthwhile.

The core features

1. Flames that visibly grow

Every flame is driven by distinct days tended, not by raw clicks or a fragile streak counter:

Flame at 7 distinct days
Blaze at 20 days
Beacon at 50 days
Phoenix at 100 days

Each stage has its own visual language: scale, glow, flicker, extra tongues, ember particles, and eventually a phoenix silhouette.

2. Consistency instead of streaks

Ember measures how many distinct days were tended in the trailing 25-day window. That means one missed day does not erase the previous month of effort.

The flame also has a live life state:

Burning: recently tended and fully bright.
Grace: a short absence dims the flame but keeps its stage.
Decayed: a longer absence lowers the visual stage temporarily.

The underlying progress is never destructively reset. The moment you log again, the flame comes back.

3. A calendar that shows a full life

The calendar makes consistency visible without turning it into a leaderboard. A day can show one flame, or several flames braided together when multiple passions were tended on the same date.

4. A journal that remembers the work

Logging a day can be as quick as choosing a mood. If you want to say more, Ember also stores an optional reflection and any AI check-in answer.

The journal then becomes a quiet record of what actually happened: the small wins, the rough days, the milestones, and the reasons you kept going.

How the AI is useful

The AI is not a generic chatbot bolted onto the side of the app. It has two focused jobs: help a new flame discover its reason, and make daily reflection easier.

AI onboarding: turning setup into a conversation

When a user creates a flame, Ember asks three questions about its origin story. Gemini rewrites each question using:

The name of the flame.
The user's previous answers.
A specific conversational intent for that step: why they started, what the passion could mean in their life, and what “good enough” success would feel like.

That makes the setup feel less like filling out metadata and more like explaining something meaningful to a supportive friend.

AI check-ins: a gentle prompt for today

When logging a day, the user can ask Ember for a personal check-in. Gemini receives the flame's name, current stage, number of tended days, and original answers. It then asks one short question that connects today's effort back to the reason the flame was lit.

The AI prompt is deliberately optional. A mood tap is always enough, and the journal field is always available without answering the AI question. The feature adds reflection when it is useful; it never becomes another obligation.

Responsible integration

Gemini runs through a server-side Convex action. The API key is stored as a Convex environment variable and never reaches the browser. If the key is missing or the request fails, Ember falls back to concise handwritten questions, so the core experience still works.

Demo

Live app: ember-two-sigma.vercel.app

bv-saketha-rama / ember

Ember — tend your passions as living flames. Next.js + Convex + Clerk + Gemini.

🔥 Ember

Tend the things you're building. Watch them burn.

Ember is a gentle passion tracker where every project, practice, or habit becomes a living flame. Show up and it grows from Spark to Phoenix. Miss a few days and it dims, but it never shames you or resets your progress.

Live demo → ember-two-sigma.vercel.app · Hackathon post → dev.to

Why Ember?

Most habit trackers are streak machines: miss one day and the number goes back to zero. Passions do not work that way. A missed Tuesday should not erase the work that came before it.

Ember measures consistency over a rolling 25-day window. Each passion has its own color, stage, origin story, and journal. The result is a calm dashboard that rewards returning instead of punishing absence.

Features

Living flames: five visible stages, from Spark to Phoenix.
Consistency over streaks: progress is based on distinct days tended in the…

View on GitHub

The main flow is:

Sign in and create a flame.
Choose its color and answer three questions about its story.
Log a mood, optional AI answer, and optional journal entry.
Watch the flame grow and celebrate stage milestones.
Review the calendar and journal whenever you want to look back.

How I Built It

Stack: Next.js 16, React 19, Convex, Clerk, Google Gemini 2.5 Flash, Tailwind CSS v4, Motion, Vitest, and Remotion.

Mechanics first

The flame behavior lives in a small pure module, convex/stages.ts. It calculates stage thresholds, the rolling consistency score, grace, decay, and the effective visual stage. Because those rules are pure, they are covered by 18 unit tests and can be reasoned about independently from the UI.

The flame is the interface

components/flame/Flame.tsx is a stage-aware animated SVG component driven by Motion. Beacon uses a dominant central flame with narrower side tongues, while Phoenix switches to a firebird silhouette. BraidedFlame.tsx keeps multiple colors visually separate when several passions share a day.

Reactive backend and private AI

Convex provides reactive queries and mutations for flames and logs. Clerk scopes the data to the signed-in user. Gemini is called only from convex/ai.ts, which keeps the secret off the client and gives the UI a predictable fallback path.

Prize Categories

Best Use of Google AI

Ember uses Google Gemini 2.5 Flash in a focused, human-centered way. It turns flame creation into a contextual three-question conversation and offers optional daily prompts that remember the user's original motivation. The AI adds warmth and continuity while leaving the user in control of what they share.

Try it

If you have something you're trying to keep alive - a project, practice, or personal goal - light a flame and see what progress feels like when it is allowed to be forgiving.

Fashioning.ai 👗: AI-powered fashion trend discovery and personalization platform

Saketha Rama — Sun, 27 Jul 2025 14:35:25 +0000

Algolia MCP Server Challenge: Ultimate User Experience

🚀 What I Built

Fashioning.ai is an AI-powered fashion trend discovery and personalization platform that leverages the Algolia MCP Server to deliver intelligent, real-time fashion insights. The application combines cutting-edge search technology with generative AI to create a comprehensive fashion intelligence ecosystem.

✨ Core Features:

Real-time Fashion Trend Discovery: Browse and search through thousands of fashion trends with lightning-fast results.

AI-Powered Trend Analysis: Get detailed insights about popularity, styling advice, and market predictions for any fashion trend.

Intelligent Search & Filtering: Advanced search capabilities with category and region filters.

Comprehensive Analytics: View trend statistics, regional preferences, and category distributions.

Contextual AI Chat: Interactive AI assistant that provides personalized fashion advice based on specific trends.

Data Enrichment Pipeline: Automated scraping and enrichment of fashion data from multiple sources.

💻 Technology Stack:

Frontend: React + TypeScript + Vite + Tailwind CSS

Backend: FastAPI (Python) + Pydantic + Uvicorn

AI Integration: Google Gemini 2.5 Pro for intelligent responses

Search Engine: Algolia MCP Server for blazing-fast search and analytics

Deployment: Google Cloud Platform (Cloud Run + Cloud Storage)

Data Sources: Vogue, Business of Fashion, Instagram trends, and more

🔗 Demo

Github Repo: github
Live URL: website
Video URL: Video

🔍 How I Utilized the Algolia MCP Server

The Algolia MCP Server is the backbone of Fashioning.ai, powering every aspect of the user experience through sophisticated search and analytics capabilities.

Multi-Index Architecture
I implemented a dual-index system:

fashion_trends: Primary index containing comprehensive fashion trend data.

fashion_news: Secondary index for fashion news and articles.

Advanced Search Implementation

Real-time Faceted Search: The application leverages Algolia's faceting capabilities to provide:

    Category Filtering: Luxury, Casual, Streetwear, Sustainable, etc.

    Regional Filtering: Global, North America, Europe, Asia-Pacific.

    Dynamic Statistics: Real-time counts and distributions.

Intelligent Data Enrichment
Built a comprehensive data pipeline that:

Scrapes fashion data from multiple premium sources.

Enriches existing Algolia records with additional metadata.

Automatically categorizes and tags content.

Updates search indexes in real-time.

Analytics & Insights
Utilized Algolia's analytics features to provide:

Total trend counts across all categories.

Regional preference distributions.

Category popularity metrics.

Search performance insights.

AI-Enhanced Search Results
Combined Algolia search results with Gemini AI to provide:

Contextual trend analysis based on search results.

Personalized styling recommendations.

Market trend predictions.

Comprehensive fashion insights.

📈 Development Process

Phase 1: Foundation Building

    Started with a robust FastAPI backend architecture.

    Implemented comprehensive Pydantic models for type safety.

    Set up React frontend with TypeScript for maintainability.

Phase 2: Algolia Integration

    Integrated Algolia MCP Server for search functionality.

    Designed efficient data models matching Algolia's capabilities.

    Implemented real-time search with faceted filtering.

Phase 3: AI Enhancement

    Added Google Gemini integration for intelligent responses.

    Created context-aware AI chat functionality.

    Built comprehensive trend analysis features.

Phase 4: Production Deployment

    Deployed to Google Cloud Platform for scalability.

    Implemented proper environment variable management.

    Optimized for performance and reliability.

🧠 What I Learned

Algolia's Power: The MCP Server's faceting and real-time search capabilities far exceed traditional database searches.

AI Integration Complexity: Combining search results with AI requires careful context management.

Production Deployment Reality: Many issues only surface in production environments.

Error Handling Importance: Graceful fallbacks are essential for user experience.

Observability: Proper logging is crucial for debugging production issues.

💡 Technical Innovations

Smart Fallback System: Implemented intelligent fallbacks that serve mock data when Algolia is unavailable, ensuring the application never completely breaks.

Context-Aware AI: Built an AI system that understands the specific fashion trend being discussed and provides relevant, actionable advice.

Real-time Data Pipeline: Created a system that can scrape, process, and index new fashion data in real-time.

Faceted Analytics: Leveraged Algolia's faceting to provide instant analytics without separate database queries.

🗺️ Future Scope & Improvement Plans

Short-term Enhancements (Next 3 months)

Advanced AI Features: Implement trend prediction algorithms, add image-based fashion analysis, create personalized style recommendations.

Enhanced Data Sources: Integrate with fashion retail APIs, add social media trend monitoring, include fashion week and runway data.

User Experience Improvements: Add user accounts and preference saving, implement trend bookmarking and collections, create shareable trend reports.

Medium-term Goals (3-6 months)

Mobile Application: React Native app with camera-based trend identification, push notifications, and offline mode.

Advanced Analytics Dashboard: Real-time trend velocity tracking, geographic trend heat maps, and influencer impact analysis.

E-commerce Integration: Shopping recommendations, price tracking, and brand partnership opportunities.

Long-term Vision (6-12 months)

AI Fashion Designer: Generate new fashion concepts, create mood boards, and predict future fashion movements.

Global Fashion Intelligence Network: Multi-language support, cultural trend analysis, and a sustainable fashion focus.

Enterprise Solutions: Offer fashion brand trend monitoring, retail inventory optimization, and market research capabilities.

💰 Potential Monetization Strategies

Premium Analytics: Advanced insights for fashion professionals.

Brand Partnerships: Sponsored trend recommendations.

API Licensing: Fashion trend data as a service.

Consulting Services: Custom fashion intelligence solutions.

🛠️ Technical Roadmap

Machine Learning Pipeline: Train custom models on fashion trend data.

Real-time Streaming: WebSocket connections for live trend updates.

Microservices Architecture: Scale individual components independently.

Global CDN: Optimize performance worldwide.

🌟 Impact & Value Proposition

For Consumers: Discover trending fashion before it hits mainstream, get personalized styling advice, and make informed fashion choices.

For Fashion Professionals: Access real-time market intelligence, identify emerging trends early, and make data-driven business decisions.

For The Industry: Democratize fashion intelligence, reduce trend forecasting costs, and accelerate innovation cycles.