Warm architectural rendering of a complex structure built from technical plans.

AI Agent engineering for operational workflows

AI Agent Engineer & Founder

I design and build production AI agents that interpret complex inputs, use tools, operate inside explicit permissions, and recover safely when work fails.

From agent architecture and evaluation to APIs, interfaces, observability, and production delivery.

AI Agent engineering through State MethodBuild a controlled AI agent around one real workflow.Start with a bounded diagnostic, then pilot and production only when the evidence supports it.Assess your workflow

I turn business workflows, documents, tools, and human review into controlled AI agent systems with observable decisions and safe actions.

01 / Founder-led businesses

Two practices. Two different kinds of work.

I operate State Method for controlled AI Agent engineering and Predella Studio for cinematic films and imagery before launch. Each has its own offer, method, and client path.

Architectural plan moving through explicit received, checked, blocked, and approved workflow states.

Business 01 / Controlled AI Agent workflow engineering

State Method

State Method diagnoses, pilots, and builds controlled AI Agent workflows with measurable outcomes, explicit authority, human approval, and recoverable failure.

Founded and led by Hugo Sequier.

The model can improvise. The workflow should not.
  • Diagnostic
  • Controlled pilot
  • Production system
Predella Studio films and imagery before launch, illustrated by the fictional Tidecourt resort, AUVRA audio product and NIVU skincare bottle.

Business 02 / Creative visualization and film studio

Predella Studio

Predella creates cinematic films and visual assets for products, hotels and places before they launch, combining creative direction, 3D and generative production.

Founded and led by Hugo Sequier.

One brand world. Many scenes.
  • Hospitality & Places
  • Products & Brands

02 / Selected work

Selected systems built for real operational work.

Production AI, data, and product systems that demonstrate the engineering foundation behind my AI Agent work.

+ More projects

Depth across AI, product, and technical systems.

Confidentiality / Sensitive client details are anonymized. Project descriptions focus on workflows, architecture, and outcomes.

03 / Expertise

AI Agent engineering from architecture to production.

The capabilities required to make agents useful, controlled, observable, and maintainable inside real workflows.

01

Agent Architecture

Agent orchestrationTool useMemory and contextMulti-agent systemsMCP
02

Control & Evaluation

Permission boundariesDeterministic checksHuman approvalEvaluation harnessesRecovery paths
03

Knowledge & Perception

RAGDocument intelligenceComputer VisionVLMsStructured extraction
04

Production Delivery

PythonFastAPITypeScriptPostgreSQLAWSObservability

04 / AI Agent delivery

From one workflow to a controlled production agent.

I build agents around explicit outcomes, tools, authority boundaries, evaluation cases, and human ownership—not open-ended autonomy.

  1. Step 01

    Bound the workflow, outcome, and human owner

  2. Step 02

    Map the inputs, knowledge, tools, and system boundaries

  3. Step 03

    Define evaluation cases, permissions, and stop conditions

  4. Step 04

    Build the smallest useful agent with deterministic controls

  5. Step 05

    Test representative, ambiguous, and known-bad cases

  6. Step 06

    Deploy with observability, recovery, and operational ownership

06 / About

AI Agent Engineer building systems that can act safely.

I specialize in production agents that combine model reasoning with tools, memory, deterministic controls, human approval, and recoverable execution.

I hold an engineering degree specialized in Data & Artificial Intelligence, with 7 years in software development, 3 years in Data Science, and 1.5 years as a freelancer.

My AI Agent work spans orchestration, tool use, memory, RAG, permission boundaries, human review, evaluation, and the APIs and interfaces needed to operate agents in production.

At BTP Consultants, I combined AI Agents with document intelligence and Computer Vision, including tool-using floorplan verification workflows that helped reduce verification time by 80%.

I’m also the founder of State Method, a controlled AI workflow engineering business, and Predella Studio, a creative visualization and film studio for products and places before launch.

7 yearssoftware development
150K+community served by a deployed AI bot
80%time reduction on agent-assisted plan checks
2founder-led businesses

07 / Contact

Build an AI Agent for a real workflow.

Start with State Method for a focused controlled-agent workflow, contact me directly for ongoing AI Agent engineering, or use Predella Studio for product launch and hospitality films and imagery.

In your first message, include your company name, business context, objective, and expected timeline.

Your details are used only to respond to this enquiry and are delivered by Resend to my email inbox. Read the Privacy Policy for retention, providers, and your rights.