I moved the Blog API's service credentials from Amplify env vars to AWS Secrets Manager — eliminating the need to redeploy after credential changes.
I diagnosed and fixed a double-text overlay bug in the markdown editor caused by CSS layering, and added a download button to the admin media gallery.
I fixed the table of contents sidebar breakpoint, anchor scroll offset for the fixed navbar, and improved IntersectionObserver scroll-spy on blog posts.
I built a media upload system using S3 presigned URLs for secure browser-to-S3 uploads and CDN-backed delivery, with custom blogimage and BlogPDF components.
I replaced ad-hoc media queries with a CSS design token system using clamp() for fluid scaling — fixing a 'grow then shrink' responsive bug in the process.
I built REST API endpoints on top of my blog's GraphQL backend so an LLM agent can programmatically create, update, and publish posts via curl.
I replaced the static markdown blog with a full CMS backed by S3, DynamoDB, AppSync GraphQL, and a custom admin UI with a rich markdown editor.
A technical deep dive into building low-latency AI response streaming on AWS - from failed REST API and SSE attempts to a production WebSocket solution with Lambda, LangGraph, and MLflow.
From fake typing effects to real streaming challenges - our journey implementing and then removing streaming responses in a Next.js + AWS Lambda + Bedrock architecture.
I integrated AWS Bedrock with Pinecone vector search to give my portfolio an AI assistant that answers questions about my work using retrieval-augmented generation.
Discover how I architected MyAI4, an AI-first streaming platform with unified intelligence, advanced family controls, and deep content analysis—leveraging serverless AWS infrastructure and multi-agent AI to deliver personalized, cross-domain recommendations and rapid service expansion.
How I built my portfolio site from the ground up with Next.js, TypeScript, and AWS Amplify — complete with a static blog, table of contents, and PDF viewer.
Returning to coding and model creation after a break from work demands.
Complete project presentation showcasing the development of a Retrieval-Augmented Generation (RAG) system using Large Language Models on Databricks platform - my capstone project for AI Data Specialist Level 7 apprenticeship.
Advanced machine learning approaches for time series forecasting and anomaly detection in financial markets, demonstrating the application of TensorFlow and deep learning techniques to protect against market manipulation and identify trading anomalies.
Advanced deep learning neural network implementation for fraud detection in financial services, demonstrating enterprise-grade machine learning solutions using TensorFlow, AWS SageMaker, and cloud computing architectures for share registrar and pension services.
Applicability of an AI Chat Assistant in resolving knowledge compartmentalization within a workplace.
Enterprise-grade automation solution developed at Equiniti combining Excel, Azure DevOps APIs, Python analytics, and Power Query to transform manual testing workflows into intelligent, automated reporting systems for large-scale project management.
Starting to write down my projects in blog / post style.