🧠 Machine Learning & AI Engineering Portfolio

by Darrell Mortalla (dmortalla)

This repository showcases a complete, production‑ready ML & AI Engineering portfolio, featuring:

• A Flagship AI Assistant (RAG + Memory + Tools + Streamlit)
• 7 fully interactive Hero Apps
• A multi‑app Suite Launcher
• 6 advanced Phase II ML Engineering Projects
• Clean architecture, modular design, and Google‑style docstrings

It is structured to demonstrate modern AI engineering depth and breadth, targeting roles in: Machine Learning Engineer, AI Engineer, Applied Scientist, and LLM Engineer.

---

⭐ 1. Flagship AI Assistant (Crown Jewel Project)

Directory: flagship-ai-assistant/
Live Demo: Add Hostinger link when deployed

A full personal AI assistant with:

🔹 Retrieval-Augmented Generation (RAG)

• Upload .txt, .md, .pdf
• TF‑IDF–based document search
• Context‑aware LLM responses

🔹 Long‑Term Memory

• JSON‑based persistent memory
• Learns preferences (e.g., “I prefer short summaries”)

🔹 Tool Calling

• Safe AST‑based calculator
• Extensible architecture for new tools

🔹 LLM‑powered Conversation

• OpenAI Chat Completions
• Graceful fallback if no API key available

🔹 Clean Streamlit UI

• Chat interface
• Document panel
• Memory viewer
• Tool call log

🔹 Docker‑ready Deployment

docker build -t flagship-ai-assistant .
docker run -p 8501:8501 flagship-ai-assistant

---

⚡ 2. Hero Apps (7 Interactive ML/AI Demos)

Directory: hero/

These apps display breadth and real‑time interactivity:

• RAG Document Advisor
• Multimodal RAG Assistant
• Semantic Search (FAISS)
• ALS Recommender
• Time Series Forecaster
• Traffic Sign Classifier (CNN)
• ECG Autoencoder Anomaly Detector

Each app includes Streamlit UI + clean backend design.

---

🛠️ 3. Suite (Multi‑App Launcher + CLI)

Directory: suite/

A unified dashboard that:

• Shows all Hero Apps
• Provides one‑click launches
• Includes a Python CLI utility

Useful as a recruiter demo hub.

---

🔬 4. Phase II ML Engineering Projects

Directory: projects/

These demonstrate deep ML knowledge, engineering practices, and classical+modern skills.

1. FastAPI Deployment (Model Serving)

fastapi-deployment/
REST API for inference, Dockerized, clean schema validation.

2. MLflow Tracking + Model Registry

mlflow-tracking/
Full experiment tracking pipeline.

3. Anomaly Ensemble (IsolationForest + LOF)

anomaly-ensemble/
Combined anomaly scoring for ops/fraud detection.

4. HAR LSTM Sequence Model

har-sequence-model/
LSTM classifier for human activity (PyTorch).

5. Transformer Time Series Forecaster

time-series-transformer/
Custom Transformer Encoder for forecasting.

6. Graph Neural Network (GCN on CORA)

graph-gnn-cora/
PyTorch Geometric GCN for node classification.

---

📁 Repository Structure

mlai-portfolio/
│
├── flagship-ai-assistant/
│
├── hero/
│
├── suite/
│
├── projects/
│
├── README.md
└── .github/workflows/

---

🎯 Recruiter Summary

This portfolio demonstrates:

• LLM integration & prompt engineering
• RAG systems & vector search
• Streamlit UI development
• API design with FastAPI
• Experiment tracking (MLflow)
• Deep learning models (LSTM, CNN, Transformers)
• Graph ML with PyTorch Geometric
• End‑to‑end deployment workflows
• Clean, well‑documented engineering practices

Together, these projects form a complete, modern ML/AI Engineering portfolio.

---

🔗 Next Steps (for Deployment)

• Add Hostinger deployment links
• Add GitHub Pages portfolio overview
• Add screenshots to enhance visual appeal
• Add CI/CD pipelines for rebuilding apps

---

© 2025 Darrell Mortalla

dmortalla.com | Machine Learning & AI Engineering Portfolio