waking up...

VIEW
Available for new projects

Kauan MendesAI Solutions Architect / Red Teaming

Transforming Human Friction into Secure, Autonomous Intelligence.

Scroll

Operational Empathy & "Calm Tech"

Most enterprise AI initiatives fail because they treat LLMs as magic endpoints, ignoring the human element. This results in unpredictable API costs, critical data leaks, and massive user resistance. Having started my career on the front lines of operations, combined with rigorous study in sociology, philosophy, and human behavior, I possess a unique "operational empathy" — technology is only valuable if it respects human cognitive load.

My architectural approach is driven by Calm Tech — the pursuit of the absolute minimum necessary to achieve maximum functionality. I build systems that adapt to human workflows, eliminating visual and cognitive noise. By enforcing strict governance, asynchronous processing, and seamless API routing, I ensure AI ecosystems operate silently and securely in the background.

 

Operational noise
by design

0%

Budget control on
AI API costs

 

Deterministic
agent workflows

LLM Governance

Strict token limits, cost caps, and guardrails that transform unpredictable AI bills into predictable investments.

Calm Tech

Systems designed to be invisible. Minimum complexity for maximum functionality — zero cognitive noise.

Operational Empathy

Deep understanding of human behavior applied to system design. High adoption, ethical AI alignment.

Blink — AI Edge-Gateway
Gateway Redis Cache LangGraph MCP Workers

Project 01

Blink

The Enterprise AI Edge-Gateway. An architectural PoC that orchestrates LLMs safely — featuring Semantic Caching (RedisVL) to slash costs, Celery for zero-blocking async orchestration, and LangGraph/MCP workers strictly isolated from enterprise databases.

LangGraphRedisVLMCPFastAPICelery
PoCDeterministic
AI budget control
WorkspaceAI — Autonomous Web Engineering
SCRAPE AGENTS DEPLOY

Project 02

WorkspaceAI

Autonomous Front-End Engineering System. A directory-based, multi-agent generative system that researches, designs, and codes complex websites from scratch in under 5 hours — featuring AI-driven animation analysis and advanced GSAP reverse-engineering.

Multi-AgentGSAPWeb Scrapingn8n
<5hWeeks of work
to rapid pipeline
n8n Automations — Workflow Orchestration
TRIGGER PROCESS ACTION BRANCH MERGE OUTPUT

Project 03

n8n Automations

Workflow Orchestration Engine. Utilizing n8n as a foundational orchestration layer to interconnect cloud APIs, trigger asynchronous agentic workflows, and manage complex data pipelines across enterprise systems — all with zero-code visual automation.

n8nAPI OrchestrationWebhooksAsync Pipelines
E2EEnd-to-end
workflow automation
01 / 03

From chaos to clarity.

A structured approach to enterprise AI architecture. Every engagement follows the same disciplined methodology — governance first, then scale.

01

Discover

I audit the existing AI landscape, identify security gaps, and map the architectural logic.

02

Design

I architect the AI edge-gateway, orchestration layers, and governance middleware.

03

Build

I engineer multi-agent workflows, red team the system, and implement guardrails.

04

Deliver

I deploy secure, deterministic systems that run silently and scale with zero data leaks.

Where architecture meets human behavior.

I am a Product-Led AI Solutions Architect who bridges the gap between complex human behavior and high-performance machine intelligence. I orchestrate AI Edge-Gateways and deterministic multi-agent workflows that aggressively reduce OpEx while neutralizing LLM vulnerabilities.

Having started my career on the front lines of operations, combined with years of rigorous study in sociology, philosophy, and human behavior, I possess a unique "operational empathy." I understand that technology is only valuable if it respects human cognitive load.

I engineer scalable, high-performance enterprise infrastructures protected by rigorous Agentic AI Red Teaming, delivering maximum functionality with zero operational noise.

Kauan Mendes

A path from operations to architecture.

2025 — Present

AI Solutions Architect & Red Teaming Specialist

Independent — Engineering AI Edge-Gateways, multi-agent orchestration (LangGraph/MCP), and Agentic Red Teaming for enterprise security. Building secure, deterministic AI infrastructure.

2025 — 2026

SDR & Automation Engineer

SaaS Company, European Market — Automated lead generation pipeline with AI semantic analysis. 4x capacity gain. Front-line experience that shaped operational empathy.

2024

People Analytics & Automation

Fortune 500 — Built automated executive reporting system achieving 99% efficiency gain. First exposure to enterprise governance requirements.

2022 — 2023

Operations & Technical Support

Multinational — Front-line operations experience. Foundation in process optimization, human behavior observation, and technical problem-solving at scale.

Technical & Cognitive Infrastructure.

Architecture & Orchestration

LangGraph
MCP
n8n
Domain-Driven Design
Multi-Agent Systems

Security & Governance

PyRIT
NeMo Guardrails
OWASP Top 10 for LLMs
Agentic Red Teaming

Engineering Stack

Python 3.12+
FastAPI
Pydantic v2
SQLAlchemy 2.0 Async
RedisVL
PostgreSQL
Celery
GSAP

Cognitive & Global Leadership

Behavioral Psychology
Portuguese — Native
English — Fluent
Spanish — Proficient

Governance, ROI, and Uncompromising Security.

0%

AI budget predictability achieved

0

Data leaks in production systems

<0h

Weeks of web dev to rapid pipeline

Agentic AI Red Teaming & LLM Governance

Adversarial Testing — Utilizing frameworks like PyRIT for automated stress-testing against prompt injections, PII exfiltration, and cache poisoning.

Dynamic Guardrails — Implementing NeMo Guardrails as a runtime defense layer to intercept malicious prompts before they reach the model.

OpEx Governance — Engineering middleware that enforces hard token limits and cost caps, transforming unpredictable AI bills into predictable infrastructure investments.

Autonomous Multi-Agent Orchestration

Stateful Orchestration — Building Master Agents (LangGraph) that securely delegate tasks to specialized sub-agents based on strict business logic.

Cloud & API Automation (n8n) — Utilizing n8n as a foundational orchestration layer to interconnect cloud APIs, triggering asynchronous agentic workflows.

Secure Tooling (MCP) — Utilizing the Model Context Protocol to run enterprise integrations as completely isolated processes — guaranteeing zero credential leakage.

Secure & Scale Your AI

Stop burning budget on redundant LLM calls.
Stop leaving enterprise data exposed.

Let's architect an AI infrastructure that respects your users, enforces governance, and scales without noise.