Experiences

About the company: A technology company focused on building advanced AI agents and the underlying infrastructure required to power autonomous, enterprise-grade AI workflows.

Designed and implemented a Kubernetes-native service stack to manage persistent authenticated browser sessions, enabling AI agents to seamlessly bypass MFA, CAPTCHAs, and complex login flows through automated Human-in-the-Loop (HITL) interventions.

Architected a highly resilient, event-driven system using NestJS, Python (FastAPI), and NATS JetStream, managing complex state machines and ensuring the reliable processing of durable events across microservices.

Engineered a robust browser automation pipeline using Playwright with dual-mode streaming (headed VNC and headless CDP), optimizing pod resource consumption and providing real-time remote debugging capabilities.

Secured the infrastructure by enforcing strict security postures, including AES-256-GCM encryption for artifacts (cookies, tokens) stored in MinIO, OAuth 2.0 Client Credentials for agent access, and multi-tenant isolation.

Accelerated the end-to-end full-stack development lifecycle by adopting an AI-native approach, heavily utilizing agentic tools like Cursor and Claude Code to deploy scalable solutions across Azure and AWS (Amazon Web Services).

About the company: Public cloud provider within the Magazine Luiza ecosystem, supporting 1,200+ external clients and workloads for a platform with 35M+ active users.

Designed and implemented an event ingestion system based on CloudEvents, processing millions of events daily using Python (FastAPI) and RabbitMQ in a fully asynchronous architecture.

Built a fault-tolerant event pipeline ensuring no data loss through idempotency, retry strategies, dead-letter queues (DLQ), and quorum queues for persistence and consistency.

Improved database scalability and performance using PostgreSQL optimizations, including indexing, partitioning, replication, and connection pooling, enabling stable operation under high load.

Developed and maintained CI/CD pipelines with GitLab, ArgoCD, Docker, and Helm, reducing deployment time from ~20 minutes to ~5 minutes and enabling consistent multi-environment releases (dev, QA, staging, production).

Implemented observability from day one using Prometheus, Grafana, FluentBit, and Sentry, significantly improving debugging speed and incident resolution through full visibility into logs, metrics, and traces.

Defined and documented the standard for asynchronous communication using RabbitMQ, enabling adoption of event-driven architecture across 11+ teams and reducing coupling between services.

About the company: B2B SaaS platform for Occupational Health and Safety (OHS), serving 15k+ users with a multi-tenant system handling complex regulatory workflows.

Led the design and development of core system modules (document management, work orders, accident investigation) using Java (Spring Boot), Node.js, and React.

Increased unit and integration test coverage from ~0% to 90% over 6 months by introducing SonarQube, reducing technical debt and improving code quality across the system.

Refactored critical database queries for high-load endpoints (EPI control workflows), replacing inefficient ORM usage with optimized SQL in MySQL/MariaDB, improving performance by 40%.

Built and maintained BI dashboards using Metabase and BigQuery, enabling data-driven decision-making for clients and internal teams.

Improved UX for thousands of non-technical users by redesigning key interfaces, reducing support overhead and improving usability of critical workflows.