Richard Kang

• AWS Certified Advanced Networking – Specialty

  AWS (last renewed/obtained: Nov 17, 2025)

  Expires Nov 17, 2028

• AWS Certified AI Practitioner

  AWS (last renewed/obtained: Nov 16, 2025)

  Expires Nov 16, 2028

• AWS Certified Cloud Practitioner

  AWS (last renewed/obtained: Nov 15, 2025)

  Expires Nov 15, 2028

• AWS Certified CloudOps Engineer Associate

  AWS (last renewed/obtained: Nov 6, 2025)

  Expires Nov 6, 2028

• AWS Certified Data Engineer - Associate

  AWS (last renewed/obtained: Nov 4, 2025)

  Expires Nov 4, 2028

• AWS Certified Developer – Associate

  AWS (last renewed/obtained: Nov 6, 2025)

  Expires Nov 6, 2028

• AWS Certified DevOps Engineer – Professional

  AWS (last renewed/obtained: Nov 16, 2023)

  Expires Nov 16, 2026

• AWS Certified Machine Learning – Specialty

  AWS (last renewed/obtained: Oct 27, 2025)

  Expires Oct 27, 2028

• AWS Certified Machine Learning Engineer - Associate

  AWS (last renewed/obtained: Nov 5, 2025)

  Expires Nov 5, 2028

• AWS Certified Security – Specialty

  AWS (last renewed/obtained: Jan 21, 2025)

  Expires Jan 21, 2028

• AWS Certified Solutions Architect - Associate

  AWS (last renewed/obtained: Oct 29, 2025)

  Expires Oct 29, 2028

• AWS Certified Solutions Architect - Professional

  AWS (last renewed/obtained: Jun 14, 2021)

  Expires May 23, 2027

• AWS Partner: Accreditation (Technical)

  AWS (last renewed/obtained: Sep 22, 2026)

  Expires Sep 22, 2029

• AWS Partner: Migration Sales Essentials (formerly Migration Ambassador)

  AWS (last renewed/obtained: Jun 16, 2024)

  Expires Jun 16, 2029

• Aug 31, 2026

  Why Token Counts Rank AI Agent Costs Backwards

  In a real AWS sandbox, the three highest-token agents were the three cheapest. Fargate compute, not model tokens, decided the per-agent cost ranking.

• Jul 14, 2025

  Conversational AI’s New Voice: Speech-to-Speech Models in Enterprise Generative AI

  A strategic perspective on implementing real-time voice-first AI in the generative era

• Mar 26, 2025

  Hosting Your LLM Model on Amazon SageMaker for AI-Assisted Coding

  Empowering Enterprise Developer Productivity with Secure, Self-Hosted AI Coding Assistants on Amazon SageMaker

• 2

  doitintl/Evaluation-Driven-Development-with-Amazon-Bedrock-AgentCore

  If you have shipped an LLM-powered agent, you have probably been asked questions like these: "Should we move to a different model?" You answered with a guess. "I changed the prompt to fix one bug. Did it break anything else?" Another guess. "A customer says the agent ignored a constraint." You can't reproduce it. These aren't problems with the LLM. They're evaluation problems. Teams can see that an agent answered, but not what it did, whether that was good, or what specifically broke when quality drops. Evaluation-Driven Development (EDD) treats agent quality the way TDD treats code correctness. Every change is measured against a fixed definition of "good", and that definition exists before the change. The workshop builds the tooling so every production agent can answer four questions with a number:

Browse more code at github.com/doitintl.