AI & LLMs

How language models work, run and get used: architectures, inference engines and agents.

Series

8 posts

Agentic AI

Building agents with Strands and Groq: tools, memory, multi-agent patterns and the protocols agents talk through.

6 posts

LLM Architectures

Learning LLM internals from Python to vLLM, following Mentiforce's roadmap and ending in a queue-scheduler project.

Nov 1, 2026·83 min read·AI & LLMs

Agentic AI · Part 8

Trip Buddy: one agent, built up in 15 lessons

One trip-planning agent grown lesson by lesson, from a SQLite table to RAG, tools, an MCP server, long-term memory and a second agent over A2A. It covers every command, error and fix from my run: uv off the PATH, an OpenAI 429 that wasn't a rate limit, a Gemini 400 caused by dropped thought signatures, and an A2A timeout that looked like an outage. There's a reproduce-it-yourself guide at the end.

Agentic AILangChainGeminiMCPA2ARAGMemoryClaude Code

Oct 31, 2026·41 min read·AI & LLMs

Agentic AI · Part 7

Evaluating, observing and governing AI agents

The course section after deployment: tracing a Strands agent with OpenTelemetry and reading the traces in the console, Jaeger, Langfuse, LangSmith and CloudWatch; scoring an agent with AgentCore Evaluations; and limiting it with Bedrock Guardrails and Cedar policies. It includes a step-by-step guide to finding the traces I couldn't see in CloudWatch and LangSmith.

Agentic AIAgentCoreObservabilityOpenTelemetryEvaluationGuardrails

Oct 31, 2026·55 min read·AI & LLMs

Agentic AI · Part 6

Shipping agents to production: Lambda, ECS Fargate and AgentCore

Moving the agents from this series off my laptop and onto AWS in five labs: AgentCore Runtime, Lambda, ECS on Fargate, AgentCore Gateway, and the MCP and A2A system from the last post on AgentCore. It covers how to choose where an agent should live, and every error I hit on Windows on the way.

Agentic AIAWSAmazon BedrockAgentCoreStrandsDeployment

Oct 30, 2026·40 min read·AI & LLMs

Agentic AI · Part 5

How agents talk: MCP for tools, A2A for other agents

A first-timer's guide to the two protocols agents talk through: MCP to reach tools, data and prompts, and A2A to reach other agents. It builds up from a protocol-only demo to Strands agents and a three-agent travel planner, and reads a Groq rate-limit error line by line on the way.

Agentic AILLMsAI EngineeringStrandsMCP

Oct 29, 2026·23 min read·AI & LLMs

Agentic AI · Part 4

Agents that work together: supervisor, agent as tool, graph and swarm

A first-timer's guide to multi-agent patterns with Strands and Groq: a supervisor with workers, an agent used as a tool, a graph with explicit routing, and a swarm that hands off work, with the lessons each one teaches.

Agentic AILLMsAI EngineeringStrands

Oct 28, 2026·40 min read·AI & LLMs

LLM Architectures · Part 6

Batching LLM Requests: Foundations for a Queue Scheduler

The foundations of my next project, a minimal LLM queue scheduler: why batching matters and what it costs, how requests arrive, how naive, dynamic and continuous batching differ, how to measure TTFT, TPOT and throughput, and the plan for building and benchmarking it.

AILLMsAI EngineeringLLM ArchitecturesProjects

Oct 27, 2026·45 min read·AI & LLMs

Agentic AI · Part 3

How AI agents remember: Mem0, LangGraph and AgentCore Memory

A first-timer's guide to agent memory: long-term facts with Mem0 and FAISS, short-term conversation state with LangGraph checkpointing, and every block of code explained.

Agentic AILLMsAI EngineeringLangGraph

Oct 27, 2026·35 min read·AI & LLMs

LLM Architectures · Part 5

vLLM Framework and Inference Engine Design

A walkthrough of what I learnt about vLLM: why it handles high concurrency so well, how PagedAttention manages KV cache memory, how its scheduler mixes prompt and decode work, and how users and applications interact with the engine.

AILLMsAI EngineeringLLM Architectures

Oct 26, 2026·18 min read·AI & LLMs

Agentic AI · Part 2

Giving an AI agent tools: from a plain function to a multi-tool assistant

How a Python function becomes a tool an agent can call, built up with a tip calculator, a three-tool sales assistant and an inventory checker, plus the lessons each one teaches.

Agentic AILLMsAI Engineering

Oct 25, 2026·18 min read·AI & LLMs

Agentic AI · Part 1

Building my first AI agent with Strands and Groq

What an agent really is, how the Strands SDK and Groq fit together, and a step-by-step first agent with a tool, a prebuilt calculator and a system prompt.

Agentic AILLMsAI Engineering

Oct 24, 2026·38 min read·AI & LLMs

LLM Architectures · Part 4

Transformer Inference Optimization Strategies

A walkthrough of what I learnt on Transformer Inference Optimization Strategies with emphasis on inference efficiency, KV cache, flash attention and quantization methods with a broader scope on learning about LLM architectures.

AILLMsAI EngineeringLLM Architectures

Oct 15, 2026·41 min read·AI & LLMs

LLM Architectures · Part 3

Transformer Architecture Principles

A walkthrough of what I learnt on Transformer architecture principles with emphasis on transformer models, attention - multi-head and cross attention, softmax functions and positional encoding methods with a broader scope on learning about LLM architectures.

AILLMsAI EngineeringLLM Architectures

Oct 10, 2026·47 min read·AI & LLMs

LLM Architectures · Part 2

Neural Network Fundamentals

A walkthrough of what I learnt on Neural network fundamentals and hardware execution with a broader scope on learning about LLM architectures.

AILLMsAI EngineeringLLM Architectures

Oct 7, 2026·40 min read·AI & LLMs

LLM Architectures · Part 1

Python Internals for LLM Engineers

A walkthrough of what I learnt during Phase 1 of learning about LLM architectures.

AILLMsAI EngineeringLLM Architectures