Apps

Explore side-projects and micro apps.

Knowledge Drip Screenshot

The Knowledge Drip

MVP

The Knowledge Drip is an AI-driven, machine-learning powered knowledge delivery platform. It continuously fetches, filters, and classifies high-quality articles using semantic search, transformer embeddings, and hybrid ranking models & then delivers curated insights to each user via SMS, tailored to their chosen interests.

As a full-stack and AI project built with FastAPI, Supabase, and Next.js, it runs a complete end-to-end pipeline, from automated content fetching and machine learning classification to database-driven personalization, user authentication, and live Twilio notifications.

  • Click “Live Demo” below to open the app.
  • Sign up with your verified phone number (Twilio-powered OTP).
  • Choose your preferred genres and delivery frequency.
  • The main engine fetches fresh articles and sends them to your inbox via SMS.
  • Rate articles (👍/👎) to improve future recommendations.

Tech Stack Overview

NLP MachineLearning InformationRetrieval SemanticSearch TransformerModels SentenceEmbeddings BM25 HybridSearch Supabase FastAPI NextJS FlyIO TwilioAPI Docker Microservices RESTfulAPIs PythonAutomation AIContentPipeline FullStackDevelopment CloudDeployment CI/CD RecommendationSystems DataPipelines ScalableArchitecture IntelligentAutomation
Live Demo Article GitHub

Note for recruiters / hiring teams: The GitHub repo for this project is public but the deployed backend is paused to keep costs down. Only the frontend demo is live. For a full end-to-end demo, contact me via the website's contact form or email me at reflections@fourthprojection.com. You can also contact me via the private email on my cv.

Market Data Platform Screenshot

Market Data Platform

Coming Soon

Market Data Platform is a personal market data system built to demonstrate a production-style cloud data engineering architecture end to end. The project intentionally serves two perspectives, On the surface, it presents a small, focused application for tracking a limited set of market symbols across owned and watchlisted assets & Underneath, it is engineered as a multi-workload data platform.

A streaming pipeline prepares data for low-latency, OLTP-style serving, while a batch pipeline produces historical aggregates and datasets designed for OLAP analytics and machine learning experimentation. Data quality is enforced through explicit validation and quarantine patterns, ensuring that only trusted data reaches downstream consumers.

Tech Stack Overview

CloudDataEngineering AWS IAM S3 Kinesis Lambda AWS Glue AWS Glue Data Catalog Athena DynamoDB EventBridge CloudWatch Terraform StreamingETL BatchETL OLTP OLAP MachineLearning TimeSeries DataQuality GreatExpectations Quarantine pattern NextJS APIs FlyIO Pytest Poetry
GitHub Live Demo (Coming Soon)

Note for recruiters / hiring teams: I am building this project intentionally from a cloud data engineering perspective first and therefore, the public UI will be deployed once the streaming and batch pipelines are fully validated end to end.