8. Data Engineering for AI Agents: Pipelines, Real-Time Sync, and Vector Search
TL;DR Getting a data pipeline right is one job. Data engineering for AI agents is a different job. The agent has to actually consume that pipeline’s output,…
7. Building Real Human-in-the-Loop with LangGraph Checkpointing
TL;DR A review queue with a button looks like human oversight. It isn’t — the automated decision has already taken effect by the time anyone opens that…
6. Compliance Infrastructure: Audit Trails, Policy-as-Code, and the Append-Only Principle
TL;DR We built three pieces of audit trail architecture for a regulated AI platform: a transactional outbox so an audit event can’t be silently lost between Postgres…
5. Red-Teaming AI: OWASP LLM Top 10 and the Probes You Actually Need
TL;DR We had heard of the OWASP LLM Top 10, and we had implemented fewer than half of it. We had PyRIT installed, and we had almost…
4. Securing Agentic AI: Authentication, Authorization, and PII
TL;DR Our five-agent banking AI platform had OPA wired into exactly one agent. The MCP server had no auth middleware. All five agents shared one API key….
3. The 57-Gap Audit — What “Done” Actually Means in Production AI
TL;DR After weeks of building a multi-agent AI platform — five agents, full pipeline, red-team harness, control UI — the system looked done. It wasn’t. A config…
2. The Engineering Platform: Orchestration, Monorepo, and the Stack Decisions
TL;DR Once you decide to build a multi-agent AI system, you face three engineering choices that will determine whether the platform is governable or just functional: what…
17. Seventeen Parts Later: Lessons From Building an Agentic Data Platform
📌 TL;DR Seventeen articles. Seventeen components. One platform that generates, governs, and serves financial data through a network of coordinated AI agents. This final article reflects on…
16. Proactive Intelligence: Building Alert Systems That Think Ahead
Series: Building an Agentic Data Platform | Part 16 of 17Reading time: ⏳ ~12 minutesTags: 🏷️ alerting alert engine notification system escalation policy Prometheus AlertManager data quality…
1. From Monolith to Multi-Agent — Why One AI Is Not Enough for Regulated Finance
TL;DR A single AI agent cannot safely handle a regulated financial decision. The problem is not capability — it is separation of concerns, auditability, and governance accountability….