Writings & Reflections

Explorations in technology, travel, and personal growth.

Google's Agent Development Kit

April 22, 2025

Practical Guide to Google's Agent Development Kit (ADK)

Explore Google's new open-source framework for building intelligent multi-agent systems. This article provides practical examples of creating, connecting, and deploying specialized agents that can collaborate to solve complex tasks, with code examples and real-world applications.

Visualizing Gradient Descent

March 11, 2025

Visualizing Gradient Descent Optimization

This article implements five gradient descent variants (SGD, Momentum, RMSProp, Adam, AdamW) in interactive visualizations. The code demonstrates convergence properties across convex, non-convex, and saddle point functions, with performance benchmarks for each method.

Object Detection Evolution

Mar 13, 2025

Object Detection Evolution: From HOG to YOLO to DETR

This article implements and benchmarks six object detection algorithms across three generations: traditional (HOG+SVM, Viola-Jones), CNN-based (RCNN, YOLO), and transformer-based (DETR, ViTDet). Each implementation includes performance metrics, inference speed comparisons, and accuracy analysis on standard datasets.

Reinforcement Learning Environments

Mar 17, 2025

Reinforcement Learning Environments: From CartPole to MuJoCo

This article implements five reinforcement learning environments with increasing complexity: CartPole (4 state dimensions), LunarLander (8 dimensions), BipedalWalker (24 dimensions), HalfCheetah (17 dimensions), and Humanoid (376 dimensions). Each implementation includes training code, reward structure analysis, and performance metrics.