• ❄️ Data Engineering & Cloud Warehousing: Focused on architecting scalable data pipelines, robust dimensional data models, and analytics workflows with Snowflake, dbt, and modern SQL.
• 🧠 Generative AI & Agentic Systems: Building autonomous agents, RAG engines, and multi-modal pipelines leveraging Google Gemini and Vertex AI Agent Engine.
• 👁️ Computer Vision & Deep Learning: Deploying real-time multi-target tracking (YOLOv5, DeepSORT), 3D landmark pose estimation (MediaPipe), and neural networks in PyTorch.
• ⚙️ Algorithms & Intelligent Systems: Designing evolutionary genetic algorithms, state machine simulations with Godot Engine, and high-performance Python backends.

Category	Technologies & Tools
Data Engineering & Cloud
Generative AI & Agents
Computer Vision & ML
Languages & Core
Tools & DataOps

Generative AI • Autonomous Agents • Multi-Modal Systems

An intelligent plant care concierge built with Google ADK, Vertex AI Reasoning Engine, RAG grounding, Firestore, and A2UI (Agent-to-User Interface). Features persistent memory banking, custom code execution sandbox, real-time weather integration, and Imagen 3 / Gemini Omni multi-modal generation.

Data Engineering • Snowflake Cloud • dbt Core • Data Quality

Production-grade ELT dimensional modeling pipeline built on Snowflake and dbt Core. Implements Kimball dimensional modeling (star schema, facts & dimensions, SCD Type 2 tracking), multi-layer transformation architecture (staging, intermediate, marts), custom Jinja macros, automated data quality tests, and CI/CD validation.

Computer Vision • Multi-Object Tracking • Deep Learning

High-throughput multi-target tracking pipeline integrating YOLOv5m for dense pedestrian detection and DeepSORT with OSNet ReID for persistent identity association across occlusions. Evaluated on the MOT20 benchmark, featuring Kalman filter 8D motion state estimation, cascaded cosine metric matching, and real-time CLI inference.

Computer Vision • Spatial Geometry • Deep Learning

Real-time computer vision pipeline utilizing MediaPipe 3D landmark extraction and OpenCV to track 3D hand poses and classify sign language alphabets with high spatial precision. Features both ANN and CNN model architectures.

Deep Learning • Natural Language Processing • Attention Mechanisms

Full vanilla Transformer sequence-to-sequence neural network implemented from mathematical first principles in PyTorch. Features 8-head multi-head self-attention, sinusoidal positional encodings, label smoothing, and Noam learning rate scheduling with conversational CLI inference.

Evolutionary Computing • Genetic Algorithms • Simulation

Autonomous gameplay agent leveraging genetic algorithms to evolve neural networks across generations in Pygame. Features real-time visualization alongside headless accelerated training modes.

Computer Vision • Convolutional Neural Networks • Interactive GUI

Interactive desktop drawing canvas application powered by a 99.65% accuracy CNN trained on MNIST with TensorFlow, featuring real-time image matrix pre-processing and dynamic digit classification in Pygame.

Machine Learning • Applied Statistics • Desktop GUI

Interactive analytical desktop application built with Python, Scikit-Learn, and Tkinter allowing users to import arbitrary datasets, configure train/test splits, and train/evaluate Linear Regression, Logistic Regression, Decision Trees, and Random Forests.