Featured Project 01
Digital Twin
A personal knowledge assistant built on my own career materials and background
content, using retrieval-augmented generation to answer questions grounded in
source documents rather than generic responses.
RAG
Document Retrieval
Tool Calling
Personal Knowledge System
- Uses my own materials as the knowledge base for more accurate, context-aware answers.
- Can call external tools such as alerts and notifications as part of a broader workflow.
- Demonstrates how AI can be personalized, useful, and grounded in trusted source material.
Featured Project 02
AI Article Writer
An AI workflow that generates written content and paired imagery from a user prompt,
combining structured prompting, tool usage, and a polished user-facing output.
LLM Workflow
Content Generation
Image Generation
Prompt Design
- Turns a simple user input into a more complete article-generation experience.
- Illustrates multi-step orchestration rather than a single isolated model call.
- Highlights product thinking around usability, output quality, and end-user value.
Featured Project 03
Netflix RAG Chatbot
A retrieval-augmented chatbot that answers questions about the Netflix Culture Memo,
using chunking, embeddings, vector search, and grounded response generation.
OpenAI API
Embeddings
ChromaDB
Gradio
- Embeds source documents and retrieves the most relevant chunks at question time.
- Uses retrieved context to reduce hallucinations and improve answer relevance.
- Shows core AI engineering patterns for building trustworthy Q&A systems.
Featured Project 04
CardioSentinel
A machine learning project that predicts heart attack risk, demonstrating end-to-end
model development with preprocessing, pipeline design, evaluation, and deployment-minded structure.
Machine Learning
Pipelines
Model Evaluation
Scikit-learn
- Built as a complete prediction workflow rather than a notebook-only experiment.
- Highlights feature preprocessing, model comparison, and practical ML implementation.
- Structured for reproducibility and scalability using modular pipelines and containerized environments.
