Home

Orchestration vs Choreography: Two Ways to Build Clinical Speech-to-Text

Published: April 16, 2026

I built clinical transcription twice, once with event-driven choreography on AWS and once with workflow orchestration in Folium. Both are “async,” but they break in different ways. Here’s the tradeoff and why it matters in healthcare.

Migrating My Portfolio to Jekyll: Why and How NEW

Published: April 7, 2026

I was embedding Notion pages in my site. Simple, but too dependent. I wanted content in Git, not locked in Notion's database. Jekyll's native GitHub Pages support made the switch effortless. Now I write markdown, commit, and it's live.

Evaluating Workflow Orchestration Frameworks NEW

Published: December 23, 2025

Built hands-on POCs comparing Azure Durable Functions, Temporal, Prefect, and Dapr for orchestrating event-driven AI pipelines. Implemented the same invoice processing workflow (PDF → parse → fan-out/fan-in → aggregate) across all frameworks for an apples-to-apples comparison. Key insights on deterministic replay, multi-service architecture patterns, and how local dev experience impacts iteration speed and debugging.

CandleWise - Part 4: Automating Deployment CI/CD

Published: August 14, 2025

Automated my entire infrastructure using Terraform for Azure. Implemented a comprehensive CI/CD pipeline with GitHub Actions that handles my complete deployment strategy - automatically deploying the Next.js frontend to Vercel and the .NET Core backend to Azure App Service.

Zero-Cloud Image Clustering

Published: May 19, 2025

This project uses lightweight models like MobileNet V3 and robust clustering algorithms (HDBSCAN and K-means) to create an efficient local image clstering solution. The implementation uses cosine similarity to compare high-dimensional image embeddings. Most importantly, it shows that modern edge devices can handle complex machine learning workflows while preserving user privacy and reducing dependency on external services.

Gesture Recognition Without the Cloud: Why On-Device AI Is a Game-Changer

Published: May 15, 2025

Implement a privacy-preserving gesture recognition directly on your device using MediaPipe and OpenCV. Process gestures locally without cloud dependencies, ensuring user privacy while maintaining high performance. The solution combines powerful ML models with robust image processing for real-time analysis, all presented through an intuitive Streamlit interface.

Forget Chatbots! Ambient Agents Work While You Sleep!

Published: January 17, 2025

MarketMind

Published: December 23, 2024

MarketMind is a Django-based web application that integrates with a language model agent to provide users with the latest news and stock prices for companies. The application uses LangChain tools to fetch data from Google Search, process it, and store it in a FAISS vector database for efficient retrieval. This project demonstrates how to build a simple yet powerful LLM-based agent that can handle real-world queries about companies.

Dragons vs Unicorns (Part III) - AI Player

Published: October 9, 2024

In the previous post we made the markup and applied some styles to our Dragons vs Unicorns game app. Now it’s time to implement the game logic in JavaScript. This part is what makes the game playful. First, we reference the html elements for accessing them more easily.

Medical Calls Analysis in AWS (Part 5) - Automating AWS Deployments with Terraform

Published: July 2, 2024

Learn how to automate AWS resource provisioning and management using Terraform. This guide covers version-controlled infrastructure, consistent deployments, and error-free configuration.

Build a Restaurant Name Generator using LangChain and Streamlit

Published: June 2, 2024

Build an application for generating unique restaurant names using Large Language Models (LLMs) and the LangChain framework. It covers setting up the environment ...

Medical Calls Analysis in AWS (Part 4) - Monitoring with AWS CloudWatch

Published: May 8, 2024 Updated: May 17, 2025

Manually tracking metrics and monitoring system performance becomes increasingly challenging and time-consuming. With numerous API calls, data processing tasks, and model invocations happening simultaneously, it's crucial to implement automated monitoring solutions. This is where a robust logging and monitoring framework becomes essential.

Medical Calls Analysis in AWS (Part 3) - Smart Summarization with Amazon Bedrock

Published: May 5, 2024 Updated: May 17, 2025

In Part 3 of my medical calls analysis project, I learned to build an automated summarization system using Amazon Bedrock's Titan LLM. I gained hands-on experience with event-driven architecture, IAM security, serverless computing, and prompt engineering.

Medical Calls Analysis in AWS (Part 2) - Audio Transcription

Published: May 1, 2024 Updated: May 17, 2025

We'll start by uploading a call segment to Amazon S3 storage. Amazon Transcribe will then automatically convert the speech to text, creating a readable transcript of the conversation. In this article, we'll set up a transcription service that monitors an S3 bucket for new audio files and creates JSON transcripts.

Medical Calls Analysis in AWS (Part 1) - Getting Started

Published: April 27, 2024

AI-powered medical call analysis is revolutionizing healthcare delivery. Using Amazon Bedrock and cloud technologies, providers can automatically convert patient conversations into searchable records, reducing administrative work while enabling data-driven decisions and improved patient care.

CandleWise - Part 3: Deploying App to Azure

Published: January 21, 2024

The blog post details the process of deploying the app to Microsoft Azure. The process involves encapsulating the application within a Docker container, creating a Docker image of the application, and using Visual Studio tools for developing with Docker containers.

CandleWise - Part 2: Integrating a third party API

Published: January 19, 2024

This article discusses the integration of real-time stock data into an application using IHttpClientFactory, a tool that manages HttpClient instances...

CandleWise - Part 1: Environment Setup

Published: January 15, 2024

CandleWise, a portfolio management app that I'm building as part of my exploration journey into .NET Core development. This series documents my learning process with Visual Studio, Docker, and deployment to Azure. In this guide, we'll cover setting up a development environment by installing Visual Studio 2022 and .NET Core 6.0 SDK, then creating an ASP.NET Core Web API project as our foundation....

From Idea to Reality: How I Created a Budget App Using React

Published: September 14, 2021

How to build a currency converter

Published: August 15, 2021

Dragons vs Unicorns (Part II) - JavaScript

Published: June 3, 2018

In the previous post we made the markup and applied some styles to our Dragons vs Unicorns game app. Now it’s time to implement the game logic in JavaScript. This part is what makes the game playful. First, we reference the html elements for accessing them more easily.

Dragons vs Unicorns (Part I) - HTML and CSS

Published: May 18, 2018

Game making is one of the most popular branches of programming. It is also a fun way to learn any new programming language while creating an interesting project. In this series, we will go step-by-step through implementing a simple Tic-Tac-Toe-like game using HTML, CSS and JavaScript. I will called it “Dragons vs Unicorns”.