Section E · Execution · Read morning-of

Day-Of Interview Tactics

Read the morning of each round.

The night before

  • Reread 00-START-HERE, 15-interview-questions, and 02-positioning-from-scratch.
  • Sleep. Caffeine recovers from one thing; sleep does not.
  • Pick a quiet room with a good camera, good light, good audio. Test it.
  • Pull up the JD in a tab; have it visible behind your camera (printed sticky if you need it).
  • Have water. Have a notepad and pen. Phone face-down, off the desk.

30 minutes before

  • Re-read your one-paragraph self-summary out loud.
  • Re-read your pre-loaded stories. Once each. Don't memorize new content now.
  • Stand up, walk for 10 minutes. Don't keep cramming.
  • Three minutes before, sit, breathe slow for 60 seconds. Smile — it changes your voice.

During — structural moves

Open every answer with a 5-second framing

Before diving in, frame the answer:

  • "Two things matter here..." (signals structure)
  • "I think about this in three buckets..." (signals structure)
  • "Short answer is X. The longer version..." (delivers value first)
  • "Depends on the risk tier — let me cover both..." (signals judgment)

This buys you thinking time and signals you're not just blurting.

When you don't know something

Never bluff

AI engineers detect bluffing instantly because they live in this space.

Patterns that work:

"I haven't worked directly with X. My closest experience is Y. Want me to reason about X from first principles, or is there a related thing you'd rather dig into?"

"I know X exists and what it's roughly for, but I haven't built with it. From what I've read it does Z — is that aligned with how you'd describe it?"

"I'd want to look at the docs before committing to an answer there. Reasoning from what I do know..."

The reframe: "I don't know" → "Here's the boundary of what I know and how I'd extend it."

When stuck on a design question

Reach for the 10-step compliance-architect sketch from 05-harnesses-and-agents:

The architect sketch

"Let me work through this systematically. First, I'd risk-tier the task — is this low/medium/high stakes? Then pick the simplest pattern that fits — workflow versus agent. List the tools, mark side-effecting ones for human approval. Define the human checkpoints. Decide what the audit log captures at each step. Plan the eval — dataset, grader, metric. Plan the failure paths. Plan the rollout — shadow mode, then HITL, then maybe autonomous for low tier. Plan ongoing monitoring. Plan the kill switch."

Even if the interviewer interrupts you mid-sketch, that opening signals architect-level thinking.

When you need a moment

"That's a good question — let me think for a second." Then actually think for 5-10 seconds. Don't fill silence with filler. Pause is fine.

When you don't understand the question

Ask. "When you say [term], do you mean [interpretation A] or [interpretation B]?" Asking for clarification is not weakness — it's signal.

When they push back on your answer

They might be testing whether you'll fold. Or they might be right. Hold your view if you have a reason; update if you don't.

"That's fair pushback. Where I'd hold ground is X — because [reason]. Where I think you've got the better of it is Y — I hadn't weighed [factor]."

That sentence shows judgment AND humility. Both win points.

Specific traps to watch for

1"What would you do if Compliance disagreed with your design?"

Do not say "I'd convince them." That's the wrong instinct.

"Compliance has the call on risk classification — that's their domain. My job is to surface the technical realities clearly: here's what works, here's the failure mode, here's what mitigation costs. Then they decide. If the call is 'too risky, don't ship,' that's a fine outcome — better than shipping regulatory exposure."

2"Would you ever let an agent file a SAR autonomously?"

No. Never. Don't hedge.

"No. SAR filing is a regulatory action with downstream consequences for the customer and the firm. The agent can draft, can recommend, can assemble evidence — humans approve and execute. Anywhere close to a regulatory filing or external customer action is high-tier and stays human-gated. Period."

3"Tell me how you'd improve our [thing]"

You don't know their thing yet. Don't fake.

"Without knowing the current architecture I'd be guessing. Can I ask a few clarifying questions and then take a stab? [Listen, then sketch]."

4"What's wrong with [their tech choice]?"

Tread carefully. Don't trash their stack on day one.

"Tradeoffs more than wrongness — [their choice] gives you X and Y, costs you Z. The places where Z matters most are [scenarios]. If those scenarios are central to the workload, I'd lean toward [alternative]. If not, [their choice] is the right call."

5"Walk me through a project end-to-end"

They want depth. Pick one project — whatever you can speak to honestly and in detail. Use STAR structure and don't sprint:

  • Situation: 30 sec
  • Task: 30 sec
  • Action: 90 sec — the meat, with specific decisions and tradeoffs
  • Result: 30 sec — what shipped and what you learned

Total ~3 minutes. They'll dig in with follow-ups.

Questions to ask them (have these ready)

Have 5-6 ready, ask the 2-3 best based on what came up:

  1. "What's the highest-stakes compliance workflow currently being considered for AI assist? What gates are non-negotiable?"
  2. "How is the team currently measuring whether an AI-drafted output is good enough before a human signs it?" (probes evals maturity)
  3. "When an AI-driven decision is later questioned by a regulator, what does your replay/audit story look like today?"
  4. "Where does n8n vs custom Python vs MCP cleanly divide responsibilities in your existing builds, and where is the boundary fuzzy?"
  5. "What's the team's biggest unsolved problem? Where would a new architect have most leverage in the first 90 days?"
  6. "How does the team handle prompt and model versioning today? Is there a registry, or is it convention?"
  7. "What does the eng-comms relationship look like — how do AI engineers and Compliance leads collaborate on a typical build?"
  8. "What's a recent thing the AI team got wrong, and what changed because of it?" (this one gets you signal on culture)
Don't

Don't ask about salary or remote in technical rounds. Save for recruiter / hiring manager.

Your closing answer

When asked "anything else you want to add?" or "any final questions?" — have one more substantive thing to say:

Honest, secure, useful — three things in one

"One thing I want to flag: I know parts of the JD describe more years of compliance experience than I have. What I bring instead is recent, focused depth on the AI engineering side — agent design, MCP, evals, error-handling, audit-aware architecture — and a serious commitment to ramping fast on the compliance domain. If the team's biggest needs are 'someone who can ship AI well in a regulated context,' I'd argue I'm closer than my CV suggests. If you're looking for deep AML or RegTech operational chops as the primary axis, I'd be honest that I'd be ramping. I'd want to know which axis matters more for this specific seat."

This does three things: acknowledges the gap, reframes around your strength, asks a smart calibration question.

State management — the meta-skill

Two AI engineering rounds back-to-back is exhausting. After round 1:

  • 5 minutes between rounds: stand up, water, breathe, reset face. Do not replay the round.
  • After both rounds: write down 5 questions they asked and 3 things you said well. That's the input for the next round if they bring you back.
  • Don't try to remember "every question they asked" — pick highlights.

If something goes badly

It will. One question will land wrong. One blank moment.

Recovery rules:

  • Name it briefly: "Let me come back to that — I want to give you a better answer than I started." Then pivot.
  • Don't dwell. The interviewer's memory is dominated by recency. Strong answer in question 9 erases mediocre answer in question 4.
  • Don't apologize twice. Once is composure; twice is anxiety leaking.

After the interview

  • Send a short thank-you email within 24 hours. Three sentences. Reference one specific thing they said. Don't restate your case.
  • Note things you'd answer differently. Adjust this folder.
  • If you advance: schedule the next round, drill again, iterate.

The one mantra

"I'm closer than I think. I have prepared seriously. Lead with concreteness."

Read that ten times before you start the call. Then go.