Positioning From Scratch

The honest premise

You're being interviewed for a senior AI engineering role. You don't have years of production AI/ML/compliance-tech work to draw on. You can't fake that, and you shouldn't try — AI engineers detect bluffing fast because the field is small and the failure modes of fake experience are predictable.

What you can do:

1. Demonstrate fast learning — the most valuable trait for someone joining a fast-moving domain.
2. Reason cleanly from first principles — show that even on unfamiliar topics, you arrive at sensible designs.
3. Lean on transferable skills — whatever your background is (PM, ops, generalist software, customer-facing tech), it has structure-thinking, communication, judgment value.
4. Show domain curiosity for compliance — they're hiring someone who'll care about audit trails and human-in-the-loop, not someone who'll resent them.

The opening posture

When you walk into the call, your default tone is calm, curious, honest, and prepared. You are not trying to be the most technically credentialed person in the room. You are trying to be the person they want to spend an hour talking with — clearly thinking, fast learning, willing to push back, willing to say "I don't know."

Repeat this before each round:

That mindset beats trying to fake seniority every single time.

How to talk about your background

The key question: how do you describe what you've done without claiming experience you don't have, and without selling yourself short?

A four-sentence template

Adjust to your actual background. The structure:

1. What you are at the core (PM / ops / engineer / generalist).
2. Why AI/compliance is interesting to you specifically — one substantive sentence, not generic enthusiasm.
3. What you've engaged with in this space recently — courses, building, reading, side projects, internal experiments.
4. What gap you're explicitly closing and how.

Example, customize freely:

That's a real answer. It does NOT claim production AI/ML expertise. It shows: thoughtfulness, deliberate skill-building, role-specific motivation, self-aware honesty about gaps.

Reframe: aptitude over experience

Hiring someone "underqualified-but-promising" is a bet on aptitude. Aptitude is observable in 60-minute conversations through:

• Vocabulary fluency — you can use the right words in the right places (this folder fixes that).
• Sound first-principles reasoning — when given a novel scenario, you arrive at a defensible architecture by reasoning, not by recall.
• Asking the right clarifying questions — knowing what to ask is half the skill.
• Disagreeing well — pushing back with a reason when you disagree with the interviewer.
• Updating well — changing your mind when they make a strong point.
• Failure-mode thinking — instinctively asking "what could go wrong here?"
• Willingness to say "I don't know" — without spiraling.

Practice each. They are not innate; they are stylistic moves you can prepare.

How to handle "have you done X?" honestly

Default template — memorize the shape:

Why this works:

• Honest gap statement — earns trust.
• What you have engaged with — shows you're not starting at zero.
• Your current model of it — demonstrates active learning, not passive consumption.
• Specific gotcha — shows you've thought about edge cases, not just the happy path.
• Invite redirection — gives them the steering wheel; respects their time.

That fifth move — inviting them to redirect — is unusually powerful. Most candidates take the floor; you giving it back signals self-awareness and seniority.

When they ask for a "real example"

Two patterns work, even without deep experience.

Pattern A: A small but real thing you actually did

Anything you've genuinely done counts: a prompt you tuned for an internal tool, a script you wrote that calls Claude, a side-project agent, an MCP server you tinkered with, a course capstone, a hackathon project. Use it.

The story is the same shape as the senior version (situation, action, decision, tradeoff, outcome) — just smaller scale. Don't apologize for the scale. "Small project, real engineering decisions" is fine.

Pattern B: A reasoning exercise

If the question genuinely doesn't fit anything you've done, redirect to thinking:

Then narrate your reasoning out loud. Better to think well in front of them than to fabricate.

Transferable skills that map to this role

Even a non-AI background carries weight. Audit your past for these:

Find your three strongest and have a sentence for each.

How to learn-fast in interview answers

When you encounter an unfamiliar concept during the interview:

1. Repeat it back to confirm: "X — is that the [concept], or are you using it differently?"
2. Acknowledge: "I haven't worked with X by that name."
3. Build from a related anchor: "It sounds adjacent to [known concept]. Is the key difference [your guess]?"
4. Let them correct you: this is a gift. They tell you what it is, you've now learned it, you can integrate it.
5. Use it correctly later in the conversation. Demonstrating that you absorbed a new concept mid-interview is itself a powerful signal.

Pre-loaded honest stories

Build 3-4 of these from your real life — even small ones. Each ~60-90 seconds. Examples to adapt:

Handling the experience question on your resume vs the JD

If they bring up the years-of-experience gap directly:

Honest. Calibrating. Confident. Lets them tell you which axis matters more — and that information is valuable to you regardless of outcome.

Vocabulary discipline

Replace "junior tells" with "senior tells":

Practice the replacements. They land.

What "winning" looks like for you

You don't need to out-AI-engineer two AI engineers. You need to be:

• The candidate they enjoyed talking to.
• Visibly thoughtful — not necessarily right about everything.
• Honest at the gap-edges, articulate elsewhere.
• Curious about their work, not performing curiosity.
• Someone whose first 90 days they can imagine.

If you walk away thinking "I told the truth, I sounded prepared, I asked good questions," you've won regardless of outcome — because the version of "yes" you'd want is one where they hire you knowing the gap, not one where you bluffed your way in.

The self-summary, fresh-start version

Memorize a version of this for "tell me about yourself":

30 seconds. Honest. Specific. Lets the rest of the conversation be a real one.

✓ Check yourself

Answer from memory before opening each one.

1. An interviewer asks whether you have run an eval program in production. You have only done manual quality reviews on a side project. Name the parts of an honest answer and what each one does for you.

   Show answer

   State the gap plainly, say what you have done, give your current model of the thing, name one gotcha you would verify, and hand the steering back. The gap statement earns trust, the rest shows you are not starting at zero, and handing back respects their time. See "have you done X?".

2. Midway through, the interviewer mentions a "four-eyes check on SAR drafts". You know human review but not that term. What do you do in the next thirty seconds, and why does it work?

   Show answer

   Repeat the term back to confirm it, say you have not used it by that name, anchor it to something you know ("is the key difference that the second reviewer sees the AI draft?"), let them correct you, then use it correctly later. It shows the fast on-ramp the role needs, and a correction becomes learning rather than a miss. See learning fast mid-interview.

3. You are asked to describe an agent you shipped. You built a small script that calls Claude with one tool for a side project. Do you tell that story or reason through a hypothetical, and what decides it?

   Show answer

   If you built something real that matches the question, tell it at its true scale, with the decisions and trade-offs; a small project with real engineering choices is fine. If nothing you have done fits, say so and reason aloud, inviting pushback. Never stretch the script into something it was not. See when they ask for a real example.

4. The interviewer raises the gap between the JD's eight years of compliance operations and your background. Do you reassure, concede, or something else, and what are the deciding factors?

   Show answer

   Calibrate. Acknowledge the gap once, state what you bring instead, and ask which axis the seat weights more. If the seat is mainly deep AML operations, you want to learn that early; if it is thoughtful AI in a regulated setting, your case is stronger than your CV suggests. Do not apologize repeatedly or claim years you lack. See the experience question.

5. You have run payment-retry incidents on call but never built an agent. Which two of your skills map to this role, and to which agent concerns?

   Show answer

   On-call and incident work maps to error handling, observability and fail-safe defaults. Retry and idempotency experience maps to agent harnesses and side-effect safety, such as preventing a duplicate filing when a step is retried. Having a sentence ready for each is better than a general claim of adjacent experience. See transferable skills.

Takeaway

• Do not fake production experience: the field is small and a bluff is easy to spot.
• An interview can show five things in an hour: fast learning, first-principles reasoning, good clarifying questions, disagreeing and updating well, and thinking in failure modes.
• For "have you done X?", state the gap, say what you have done, give your model of X, name a gotcha, and hand the floor back.
• Any real project counts at any size; if none fits the question, reason aloud instead of inventing one.
• For an unfamiliar term, repeat it back, anchor it to something you know, accept the correction and use the term correctly later.
• When the years-of-experience gap comes up, say it once, redirect to what you bring and ask which axis matters more.

Honest positioning only works on a solid technical base, so the next chapter covers the model and API fundamentals everything else uses. → AI Development Fundamentals