The AI-Augmented Engineer
Expectations for engineers operating in the Velocity Framework.
The AI-Augmented Engineer is not a junior developer who lets AI "do the work." They are a senior operator who uses AI to extend their reach while owning correctness, design, and quality.
Outcomes We Expect
We aim for at least 30%+ productivity gain over a traditional senior developer, without regressing on quality or security. That gain shows up as:
- Faster understanding of unfamiliar codebases.
- Smaller, more frequent, and better-structured changes.
- Higher test coverage for the same or less effort.
Core Behaviors
An AI-augmented engineer consistently:
- Clarifies with AI before coding
Uses AI to interrogate requirements, surface edge cases, and remove ambiguity (Step 1 of the workflow). - Thinks in plans, not prompts
Treats AI as an executor of a concrete plan tied to files, functions, and tests—not as a magic box that "just fixes it". - Maintains context hygiene
Resets or narrows context when switching tasks, avoids giant, unfocused chats, and keeps specs/plans in files instead of buried in conversations. - Reviews every change
Reads diffs carefully, asks "Does this match the spec?" and "Is this consistent with our rules?", and is willing to discard AI output that feels wrong.
Requirements for the Role
To operate effectively in this model, you must:
Master the Fundamentals
You cannot safely review AI-generated code if you do not understand the underlying systems.
- You spot security vulnerabilities, race conditions, and performance traps.
- You recognize when the AI is inventing types, APIs, or behaviors that do not exist.
- You know when to simplify an approach instead of layering on more code.
Treat Prompting as an Engineering Skill
Prompting is how you program the collaborator, not a bag of tricks.
- Provide explicit goals, constraints, and relevant files.
- Iterate based on the diff and tests, not vibes.
- Encode recurring instructions in project rules/config instead of repeating them manually.
Embrace the Reviewer Mindset
Your primary job is to make sure the final state of the system is correct and maintainable.
- You are comfortable editing and simplifying AI output instead of chasing a perfect one-shot answer.
- You use AI as a sounding board for refactors, trade-offs, and test strategies.
- You accept that "the AI wrote it" is never a valid excuse for a bug.
Career Trajectory
This model flattens the learning curve for syntax but raises the bar on architecture, communication, and judgment.
We value engineers who can:
- Design systems and workflows that AI can operate in safely.
- Communicate intent clearly to both humans and tools.
- Continuously refine rules, specs, and processes based on real-world feedback.
The rest of the framework describes the environment you work in: pods, governance, AI stack, and the 6-step loop that ties it all together.