[ agentic.engineer ]

Agentic
Engineer

Builds AI agent systems. Works like one too.

There's an engineer who designs agent systems, pipelines of AI that reason and act on their own. They work this way too. Take a goal, figure out the path, ship it. Do it again. The way they build software looks a lot like the software itself.

// the shift

The rules changed.

Software engineering was already hard. Then AI became capable. Some engineers absorbed this and started operating at a different layer entirely.

Before
  • Implement
  • Write line by line
  • Sprint cycles
  • Individual output
  • Wait for requirements
  • Own the code
After
  • Orchestrate
  • Direct agents at scale
  • Ship in continuous loops
  • Multiplied output
  • Define the goal, move
  • Own the outcomes
// dual nature

Two things at once.

01

The Craft

Agentic Engineering

Designing systems where AI does real work on its own. Agent pipelines that plan and execute. Tool-calling loops that gather information and produce results. Multiple models coordinating toward a single goal. Workflows that run without someone babysitting every step.

  • Multi-agent orchestration
  • Tool-calling & MCP systems
  • End-to-end workflow design
  • LLM pipeline architecture
02

The Mindset

Being Agentic

How you operate, not what you know. Agentic engineers take a goal and move. They break down problems on their own, pick the right tools, adjust as they go. Nobody needs to check on them.

  • Self-directed within scope
  • Initiative without permission
  • Goal-in, results-out
  • Self-correcting loops
// what defines them

The profile.

01

Orchestrates intelligence

Designs systems where AI does real work. Thinks in pipelines and feedback loops, not individual prompts.

02

Thinks in systems

Inputs, agents, tools, outputs, and the edges between them. Sees the whole graph.

03

Ships in loops

Do the thing, see what happens, adjust. Work actually finishes instead of piling up in a backlog.

04

Delegates to agents

Hands off research, validation, and execution to AI. Stays accountable for what comes out.

05

Runs on minimal oversight

Give them a goal and context. They figure out the rest.

06

Builds the tools that build

MCP servers, agent harnesses, custom tooling. The stack is something you shape.