Current Projects & Experiments
Healthcare AI Platform | Azure AI, ASP.NET Core, RAG
- Designed and implemented an intelligent chat interface using Azure AI Agents and Foundry, enabling users to interact with healthcare documents via natural language.
- Developed secure, scalable ASP.NET Core APIs for uploading, parsing, embedding, and indexing healthcare documents into Azure AI Search.
- Integrated Retrieval-Augmented Generation (RAG) pipelines to provide real-time, contextual responses grounded in user-uploaded medical documents.
- Engineered a document ingestion workflow using Azure Blob Storage and Azure OpenAI Embeddings, supporting fast and accurate semantic search.
- Collaborated cross-functionally with product managers and data teams to translate healthcare domain requirements into scalable AI-driven solutions.
- Optimized full-stack performance and reliability, using C# backend APIs and Angular front-end where applicable.
- Enabled analytics and monitoring for AI interactions using Azure Application Insights, improving chatbot reliability and user trust.
- Implemented Single Sign-On server to enhance secure access across platforms.
- Developed microservice architecture for scalable and efficient application deployment.
- Utilized OAuth and OpenID Connect for robust authentication processes.
- Deployed applications using Docker containers for consistent environment replication.
- Configured Azure Container Registry to manage container images effectively.
- Leveraged Azure Kubernetes Service for streamlined workload orchestration.
- Converted real-time data from database collections into standardized HL7 FHIR resources, enabling interoperability with external healthcare systems and compliance with healthcare data exchange protocols.
EdTech AI Platform | GPT-4 via Azure OpenAI, ChromaDB, Nodejs, Angular, Charts.js
- Designed AI workflows for generating curriculum-aligned content using GPT and domain knowledge bases.
- Built intelligent “Second Brain” prototypes that summarized key topics for science and history lessons.
- Developed dynamic, adaptive learning modules that modified difficulty based on student interaction.
- Collaborated with educators to fine-tune AI outputs for pedagogical alignment and engagement.
- Delivered structured JSON/graph-based content maps to support platform logic and context management.
EdTech AI-Powered Adaptive Learning Platform | Solo Developer, Azure AI, ASP.NET Core
- Developed a full-stack web application to help schools create, distribute, and analyze student surveys.
- Leveraged AI/NLP (e.g., OpenAI API) to automatically generate survey templates from brief descriptions or prompts, reducing manual effort by up to 80%.
- Supported both single- and multiple-choice questions with predefined impact indicators for automated scoring and evaluation.
- Implemented real-time response tracking and auto-summarization of survey results using AI, enabling faster data-driven decisions.
- Built interactive dashboards with graphs (e.g., using Chart.js) to visualize trends, comparisons, and key insights.
- Enabled personalized invite links for students, ensuring secure and organized participation tracking.
- Designed a scalable backend with Node.js and MongoDB and a responsive frontend using Angular.
AI-Powered Visual Learning Platform | Solo Developer, Angular, Node.js, Ionic, OpenAI, AWS
- Developed cross-platform frontends using Angular (web) and Ionic (mobile).
- Engineered a backend in Node.js to handle PDF parsing, AI job queues (BullMQ), and content orchestration.
- Integrated OpenAI GPT-4 for chapter summarization, concept extraction, and quiz generation.
- Used DALL·E 3 for image creation, ElevenLabs for narration, and FFmpeg for dynamic video generation.
- Designed and deployed a parallel processing pipeline to handle multi-chapter uploads with Redis queueing and AWS S3 storage.
- Architected system to support future semantic search using vector databases like Pinecone