Interview Prep Hub

AI Agents in Compliance

A study guide for AI engineering and solutions-architect interviews focused on agentic AI in regulated finance — AML, KYC, sanctions, RegTech. 22 chapters with sticky navigation, syntax-highlighted code, copy buttons, drill modes, and dark mode.

Build the agent that calls the sanctions APIagent(harness)MCP serversanctionsKYCAMLevals← every calllogged + scoredRAG overregsch 04–09: you build it · JWT + HTTP transport · RegTech
If you only have an hour

Open 00-START-HERE first. Then drill 15-interview-questions with drill mode on. Reread 16-day-of the morning of.

Drill: 28 Questions Hide-show answers · timer · progress Drill: 11 Coding Problems Multiple approaches · drill mode Hands-On: Build an MCP 45 min · Python · Claude Desktop
Study pathsPick the plan that fits the time you have
7+ days
  • Day 1: 01, 02 (orient) → 03 (AI dev fundamentals)
  • Day 2: 04 (MCP deep dive), then do the build tutorial
  • Day 3: 05, 06 (harnesses, RAG)
  • Day 4: 07, 08, 09 (evals, error handling, audit)
  • Day 5: 10, 11 (coding — drill problems on a timer)
  • Day 6: 12, 13 (pipelines, deployment) + try one MCP stretch (04a-d)
  • Day 7: Drill 15 in HTML drill mode. Read 14 and 16. Sleep.
2-3 days

01, 02, 03, 04 (deep-dive + build), 05, 07, 08, 11 (do the 3 named problems), 15 (drill), 16. Skim everything else.

< 24 hours

01, 02, 11 (do the 3 named problems on paper), 15 (drill all 28), 16. Skim 04, 05, 07 headings only.

Section A · Orient
00 Start Here Master index, study schedule, key reframe. Read first. 01 The Role Decoded JD breakdown, tech stack purpose, compliance constraints, what to ask them. 02 Positioning From Scratch Mindset before content — how to interview honestly when starting fresh on the technical side.
Section B · AI Engineering Concepts
03 AI Development Fundamentals Models, prompts, tool use, structured output, prompt caching, prompt injection, observability. 04 MCP Deep Dive MCP protocol primitives, transports, auth. Plus 8 drillable interview probes with strong answers. CriticalDrill mode 04 · Hands-on Build an MCP Server 45-minute tutorial. From pip install to a working agent in Claude Desktop. Compliance-flavored. Plus 4 stretch goals. Hands-on · 45 min 05 Harnesses & Agents "Harness" disambiguation, the agent loop, Anthropic's Building Effective Agents patterns, frameworks compared. 06 RAG Applied Production RAG — chunking, hybrid search, reranking, agentic RAG, Ragas evals, 5 failure modes, drillable probes. Drill mode 07 Evals Eval frameworks, LLM-as-judge, eval-driven development, agent evals. Critical 08 Error Handling for AI AI-specific failure modes, retry/repair patterns, idempotency, fail-safe defaults. Critical 09 Audit Trails Event schemas, versioning, GDPR vs retention, model risk management. Critical
Section B · Stretch tutorials (continue from 04)
04a · 25 min Add an MCP Resource Expose authoritative documents the model can reference. Tools vs resources — the distinction matters. Hands-on 04b · 25 min Add a Prompt Template Ship a one-click compliance workflow as a slash command. Evaluator-optimizer pattern materialized. Hands-on 04c · 30 min Hit a Real Sanctions API Replace the mock with OpenSanctions. Timeout, fallback, caching, structured errors — production tool design. Hands-on 04d · 30 min HTTP Transport + JWT Auth Run the server over Streamable HTTP. Add JWT-scoped auth. Now you can answer "stdio vs HTTP, when?" with conviction. Hands-on
Section C · Coding (DSA)
10 Coding Fundamentals Pattern menu, arrays, hashing, strings, sorting, DFS/BFS, two pointers, sliding window, prefix sum, Big-O cheat sheet. 11 Coding Problems Worked Out The three problems your notes called out — Longest Common Prefix, Valid Anagram, Subarray Sum Equals K — plus 8 companions. Drill mode hides solutions; multiple approaches per problem; progress tracker. Drill mode
Section D · Production / Cloud
12 Data Pipelines ETL/ELT, batch vs streaming, Airflow/Spark/Kafka, RAG indexing pipelines, AWS vs GCP shapes. 13 Model Deployment AWS (Bedrock, SageMaker, Lambda) and GCP (Vertex AI, Cloud Run). Two reference architectures. CI/CD for AI.
Section E · Reference & Execution
14 Compliance Context AML / KYC / SAR / EDD vocabulary, jurisdictions, crypto-exchange-specific wrinkles. 15 Practice Interview Questions 28 Q&A across 8 sections. Drill mode hides answers by default. Section nav, per-question practice tracker, progress saved locally. Drill mode 16 Day-Of Tactics Structural moves, traps to watch, recovery patterns, questions to ask them, closing statement. Reread morning of.