Tag: Architecture Patterns
Posts with this tag
- Cloudflare's Open Agentic Internet Blueprint: Readable, Discoverable, Callable, and Payable Web
An architectural deep dive into Cloudflare's proposed Agentic Internet framework, spanning Web Bot Auth, Markdown for Agents, WebMCP browser tool exposure, and x402 micro-payments.
- Kimi-K3 Enterprise On-Premises Deployment TCO: GPU Topologies, Power Demands, and Infrastructure Realities for a 2.8T MoE
A comprehensive breakdown of GPU memory topologies, server hardware budgets, 3-phase power, liquid cooling requirements, and a 5-stage TCO evaluation framework for deploying Moonshot's Kimi-K3 2.8T MoE on-premises.
- Running Kimi-K3 on 80 RTX 5090 GPUs: Hardware Ledger and Engineering Trade-Offs of Consumer GPU Clustering
A deep dive into deploying Moonshot Kimi-K3 2.8T MoE on 80 consumer RTX 5090 GPUs: why 44 GPUs fall short, node topologies, 25GbE networking, TCO ledger, and power constraints.
- Analyzing OpenAI GPT-5.6: Frontier Intelligence and the Architectural Shift to 'Intelligence per Token'
A comprehensive deep dive into OpenAI's official GPT-5.6 release: Sol/Terra/Luna tiering, pricing matrices, Terminal-Bench 2.1 data, Triton kernels, and Agentic Harness routing.
- 2025/2026 Modern Large Language Model Architecture Deep Dive: From DeepSeek V3 to Llama 4
Exploring the key architectural innovations behind DeepSeek V3, Llama 4, Gemma 3, and Kimi K2, including MLA, MoE, and sliding window attention engineering practices.
- AI Agent Guide: Architecture, Tools, Evaluation, and Enterprise Delivery
A practical guide to agents versus workflows, single- and multi-agent architecture, tools and MCP, state and memory, evaluation, security, and the path from PoC to production.
- Enterprise RAG Guide: Retrieval Architecture, Evaluation, and Production Delivery
A practical framework for enterprise RAG data pipelines, hybrid search, reranking, GraphRAG, agentic RAG, evaluation, access governance, failure diagnosis, and operations.
- Building Enterprise-Grade AI Agents on AWS: Bedrock AgentCore and HoyaBit's Journey from POC to Production
A summary of the AWS × HoyaBit session: the four major pain points of Enterprise Agentic AI, the Amazon Bedrock AgentCore (Runtime/Memory/Gateway/Governance) technology stack, and how Taiwan's FSC-compliant exchange pushed their voice-trading Agent and enterprise brain platform into production.
- Migrating from K8s to Amazon GameLift: Palworld Multiplayer Server Architecture in Practice
Summary of the AWS × Pocketpair session: Amazon GameLift's managed value and cost optimization, and how Palworld achieved the migration of persistent worlds from Kubernetes to GameLift using state externalization, lifecycle adapters, exactly-one world instances, and existing monitoring assets. Also, a breakdown of three major failure cases: Terraform Drift, false-healthy Fleets, and Ping Beacons.
- Hardware-Software Co-Optimization in Practice: How AWS Custom Silicon × Tomofun Furbo Slashed AI Inference Costs by 82%
A summary of the talk by AWS's Howard and Tomofun (Furbo)'s Ricky, supplemented with the official AWS tech blog: from the vertical integration of Trainium/Inferentia/Graviton/Nitro, to Furbo porting BLIP to Inferentia 2, two-tier Auto Scaling, and AMI cold start optimization, successfully saving 81.6%–83% in AI computing costs.
- From Multi-Agent Architecture to Recruiting an AI Employee in Two Minutes: AWS × Super 8 (ORRA) Enterprise Implementation
A summary of the AWS and Super 8 (ORRA) session: the single agent decision loop, three major multi-agent orchestration patterns (Graph/Swarm/Workflow), A2A communication, Amazon Bedrock AgentCore core components, and how ORRA allows business users to build and deploy AI employees in two minutes using Job Descriptions.
- Decompose with Care: Architecture Patterns, Engineering Disciplines, and Hard Lessons in Banking System Modernization
A comprehensive summary of the AWS ProServ senior consultant's presentation 'Decompose with Care': How a leading Southeast Asian bank modernized its omni-channel monolithic platform serving 20 million active users to AWS cloud-native microservices with zero downtime. Covers four major challenges, Strangler Fig pattern, 3-Tier Facade, Contract-First / Mock-First approaches, and engineering guardrails in the AI era.
- Securely Implementing Multi-tenant AI Agents on AWS EKS: From Sandbox Isolation to BitoClaw Real-world Practice
A summary of the salon sharing by AWS Solutions Architect HC and Bito Group Operations Manager Michael: OWASP LLM Top 10 security red lines, runC / gVisor / Kata sandbox comparison, Kubernetes multi-tenant isolation levels, and how BitoClaw builds a compliant, low-cost AI Agent platform using EKS, KEDA Scale-to-Zero, Pod Identity, and Network Policy.
- From Alexa for Shopping to Agentic Commerce: How Amazon × TapPay Make AI Actually Checkout for You
A summary of the talk by Amazon and TapPay VP Joseph: How Alexa for Shopping (formerly Rufus) achieved $12B revenue with a single-agent architecture, the difference between Agentic Commerce and traditional shopping guides, and AI autonomous shopping safety guardrails like one-time virtual cards, intent validation, and limit management.
- Google ADK 2.0: Workflow Graphs, Task Collaboration, and HITL Boundaries
A source-grounded analysis of how ADK 2.0 separates deterministic routing from LLM reasoning, with production boundaries for workflows, tasks, human approval, and durable state.
- Enterprise AI Agent Security: Threat Model, Control Plane, and Rollout Checklist
Build a testable defense-in-depth architecture around prompt injection, tool authorization, exfiltration, memory and identity, supply chain, and observability boundaries.
- MCP 2026-07-28: Stateless Core, Tasks, Apps, and Migration Decisions
A practical reading of the Model Context Protocol 2026-07-28 breaking changes, official Tasks and Apps extensions, and the compatibility and security work required for enterprise migration.
- Financial-Grade Enterprise Agentic AI Architecture Design: From Demo to Agentic Operating System
AI Summit Recap: Enterprise AI Control Plane, 15+ Agents responsibility breakdown, 4-stage runtime workflow for wealth managers, 3-layer security boundaries, LLM-as-a-Judge quality governance, and E·P·J·T reusable capability foundation.
- Financial AI Engineering Platform Engineering: Building Operational Agentic AI with Cloud-Native Architecture
Summary of my Cloud Summit sharing: Three lifelines for financial AI deployment, why PoCs get stuck, the three-tier architecture of Cloud Native AI Runtime, MCP tool governance, Hybrid Search and Agentic RAG, and why accuracy is a workflow property rather than a model feature.
- Martin Fowler on Harness: Building Trust in Coding Agents with Control Loops
A deep dive into Thoughtworks' analysis: guides/sensors, computational/inferential, three types of regulation, and the behavior harness gap. By comparing with OpenAI's practical experience and common Agent failure modes, we present an actionable Harness checklist.
- Ignorance.ai Playbook: The Converging Harness Practices of OpenAI, Stripe, and OpenClaw
An in-depth review of the February 2026 horizontal roundup: an engineer's work splitting into 'building the environment' and 'managing Agents', architecture as guardrails, tools as feedback, AGENTS.md as system records, and the separation of planning and execution.
- Harness Design for Long-Running AI Engineering: Generation, Evaluation, and Verification Chains
Based on Anthropic's 'Harness design for long-running application development': Improving the reliability and controllability of long-running tasks through generator-evaluator separation, external evaluation, and QA contracts.
- Long-Running Agent Harnesses: Handoffs, Verification, and Recovery
A source-grounded analysis of Anthropic's initializer, progress artifacts, feature inventory, Git checkpoints, and end-to-end verification pattern for work spanning context windows.
- Building Effective AI Agents: An Overview of Architecture Patterns and Implementation Strategies
Adapted from Anthropic's 'Building Effective AI Agents': From single agents to multi-agent collaboration, common architectural patterns, workflow design, and how to choose the right architecture based on control requirements, problem complexity, and resources.
For speaking invitations, internal engineering sessions, or architecture exchange, see the topics and public work I can bring into the conversation.
Speaking & contact