✦ Principal Architect, Agentic AI & Technical Strategy

I design software architectures that humans and AI agents can actually build on.

Over the past 20 years, I've scaled distributed platforms at eMAG, i6 Group, and Printify. Today, I combine executive technical strategy with spec-driven agentic development environments, machine-readable architecture systems, and production AI products that turn engineering organizations into high-velocity multipliers.

Explore Advisory & Mentorship → View Shipped Systems ↓
Dan Gurgui, Principal Architect, Agentic AI & Technical Strategy
20+ YRS
Production Architecture

From high-traffic e-commerce monoliths to cloud-native microservices & AI pipelines

10+ TEAMS
Cross-Org Architecture

Led architecture and reliability governance across 10+ engineering teams at Printify

2 MO → 1 WK
Integration Velocity

Cut enterprise airport onboarding time at i6 Group through platform architecture

60 TOOLS
Mapped Live on Tectonic

Maintaining the industry's opinionated Wardley Map of AI engineering practices

✦ The Architectural Shift

AI didn't kill software architecture. It made bad architecture fatal.

Giving 50 engineers an LLM coding assistant without architectural guardrails doesn't produce 10x velocity: it produces 10x technical debt.

As code generation becomes a commodity, the bottleneck in software engineering has inverted: from code-first to requirements- and specification-first.

For AI agents and human teams to collaborate safely at enterprise scale, your architecture cannot live in outdated Confluence pages. It must be explicit, modular, and machine-readable: enforced through structured specifications, domain boundaries, automated hooks, and property-based verification.

I don't build prompt demos. I architect governed systems that ship to production.

🌀

Ad-Hoc Vibe Coding

  • 50 engineers × 50 prompts = 50 divergent conventions
  • Architecture lives in tribal knowledge and stale Confluence pages
  • No evaluation loops: quality is anecdotal
  • Tech debt compounds faster than anyone can review it
🧭

Governed Agentic Architecture

  • Specs, steering files and hooks bound every agent to the codebase's rules
  • Machine-readable domain boundaries and architecture registries
  • Property-based verification and automated evaluation gates
  • Velocity that compounds: each agent run strengthens the system
✦ The Reference Architecture

The 4 layers of an agent-ready engineering organization

L1

Machine-Readable Context

AGENTS.md, domain canvases, and steering files that encode module boundaries, ubiquitous language, and forbidden patterns directly in the repo.

L2

Spec-First Work Units

Formal, machine-verifiable specifications defining requirements, invariants, and API contracts before a single line of code is generated.

L3

Scoped Multi-Agent Orchestration

Specialized agents for requirements, design, implementation, and testing, wired to real system state via Model Context Protocol (MCP).

L4

Deterministic Verification Gates

Automated editor hooks, contract tests, and property-based evaluation loops that catch hallucinations before human review.

✦ Proof of Work

Systems in Production

🗺️ Live Platform · Public Artifact

tectonic.sh: The AI Engineering Wardley Map

An interactive, continuously updated Wardley Map tracking 60 AI engineering tools, agent frameworks, and development practices from Genesis to Commodity. Built to help CTOs decide what to adopt, what to wait on, and what is already commoditized.

Wardley Mapping · AI Strategy · Eval-Driven Dev · Agent Infrastructure

Architecture & Methodology ↓
  • Intake Rule: tools and practices only move or enter the map after two independent production signals within a single week (for example, an enterprise adoption benchmark plus a major release or security incident).
  • Data Model: structured across two Wardley axes, Value-Chain Visibility (Infrastructure → Platform → Tooling → Practices → Business Capabilities) and Evolution (Genesis → Custom-Built → Product → Commodity).
Explore tectonic.sh ↗
🏗️ Enterprise Architecture · B2B

Enterprise Spec-Driven Agentic Platform

Architected an AI-native software delivery environment for a 25-engineer organization. Replaced ad-hoc prompting with multi-agent orchestration across requirements generation, system design specs, code generation, and automated testing: governed by steering files and MCP integrations.

AWS Kiro · Amazon Bedrock · MCP · NestJS · Angular · Spec-Driven Dev

Architecture & Key Trade-Offs ↓
  • The Bottleneck: 25 engineers prompting LLMs ad-hoc across NestJS and Angular produced divergent patterns and heavy PR review rework.
  • The Architecture: spec-first workflow built on AWS Kiro and Amazon Bedrock. Every feature begins as a machine-checked specification; agents are constrained by repository steering files, custom MCP servers exposing internal schemas, and automated property-based test hooks on file save.
  • Key Trade-Off: trading 20 minutes of upfront specification rigor for the elimination of multi-day PR rewrites.
🧬 Principal Engineering · Scale

Living Architecture & Domain Registry (Printify)

Led cross-organizational architecture across 10+ teams during a critical merger initiative. Designed machine-readable domain canvases and automated pipelines that decompose RFCs and ADRs into bounded architectural changes: improving reliability and lifting marketplace GMV.

DDD · Event-Driven Architecture · AWS · TypeScript · SLO Governance

Architecture & Key Trade-Offs ↓
  • The Bottleneck: static RFCs and ADRs drifted from reality across 10+ teams during a complex post-merger platform integration.
  • The Architecture: machine-readable domain canvases and event-driven boundary definitions that decompose RFCs and ADRs into verifiable architectural changes, paired with strict SLO governance across supply and fulfilment flows.
  • Key Trade-Off: enforcing asynchronous, event-driven contracts between pillars over short-term synchronous convenience.
🚀 Founder & AI Architect · Live SaaS

Fotogen & Dronicon: Production AI Ventures

Built two end-to-end AI platforms from scratch:

  • Fotogen (fotogen.ro): multi-agent e-commerce visual pipeline: shop profiling → prompt orchestration → Gemini image synthesis → automated hallucination detection.
  • Dronicon (dronicons.com): 3-layer Medallion data pipeline using GPT-4o structured extraction and Qdrant vector similarity for catalog entity resolution.

Gemini · GPT-4o · Qdrant Vector DB · NestJS · Pulumi IaC

Architecture & Pipelines ↓
  • Fotogen pipeline: multi-agent orchestration (Shop Profiler → Style & Prompt Planner → Gemini Visual Synthesizer → Hallucination Auditor Agent that flags defective renders for human review before publishing).
  • Dronicon pipeline: 3-layer Medallion architecture (Bronze raw scrapes → Silver GPT-4o structured spec extraction → Gold Qdrant vector similarity entity resolution across fragmented vendor catalogs).
Earlier career highlights: i6 Group & eMAG
  • i6 Group: led SOC2 compliance across 20 engineers; genetic-algorithm route optimization reducing fuel truck costs and time by 90%; multi-product aviation integration platform.
  • eMAG: ML product matching across millions of SKUs; monolith-to-microservices migration on Romania's largest e-commerce platform.
✦ How It Works

From first call to roadmap

01

Share Your Context

Tell me what you're building: stack, roadmap, and where it hurts. No slide deck needed.

02

Audit & Diagnose

Deep dive into your systems, specs, and AI adoption: against real code, not templates.

03

Prioritized Roadmap

A written, sequenced plan: architecture, reliability, and governed agentic delivery.

04

Executive Readout

90 minutes with your leadership team: decisions you can act on the same week.

✦ Engagements

How we can work together

🏛️ Advisory & Architecture Consulting

The Agentic Engineering & Architecture Audit

For CTOs, VPs of Engineering & founders. Move your engineering organization from scattered AI experiments to governed, repeatable velocity, or untangle the distributed systems bottlenecks holding back your roadmap.

  • Spec-Driven AI Delivery: multi-agent workflows tailored to your codebase
  • Architecture & Scalability Audits: decoupling monoliths, DDD boundaries
  • Technical Strategy & Build-vs-Buy Roadmapping: your stack plotted on the Tectonic Wardley Map

2-week fixed-scope audit · fractional principal architect retainers

See the Advisory Offer →
🧗 1-on-1 Mentorship

Architecture & AI Altitude Sessions

For senior, staff & principal engineers. AI is rapidly commoditizing pure code execution: move up the value chain into systems architecture, staff/principal influence, and AI-native technical leadership.

  • Personal Value-Chain Mapping: which skills are commoditizing under you
  • Operating at Principal Altitude: cross-org RFCs, ADRs, executive alignment
  • Hands-On Agentic Architecture: production multi-agent workflows & evals

90-min diagnostic session · quarterly mentorship track

See the Mentorship Offer →
✦ Writing & Publications

Architecture for Growth

Essays on spec-driven AI engineering, distributed systems, and technical strategy.

View All Articles →
✦ Background

Built on 20 years of shipping under pressure

I started writing production code in 2007, progressing from early web platforms to leading the monolith-to-microservices transformation and ML catalog deduplication at eMAG, building aviation platforms and leading SOC2 at i6 Group, and serving as Solutions Architect across 10+ teams at Printify (2023–2026). Since 2024 I lead my independent architecture and agentic AI practice at gurgui.com. Architecture only matters if it improves business outcomes and human leverage.

Full Story & Credentials →
✦ Let's Talk

Ready to make your architecture agent-ready?

Book a 30-minute call. Tell me what you're building: I'll tell you where the highest-leverage moves are.

Book a Discovery Call →