Open to opportunities

Afonso Spezia

@afonsospezia

Message

Backend engineer with 6 years building resilient microservices and LLM agent infrastructure in production.

Brazil

Message

What I'm looking for

I’m looking to build resilient backend services and LLM agent infrastructure that automate real workflows—owning architecture, observability, and reliability end-to-end in a production-minded engineering culture.

I’m a backend-focused Computer Engineer with 6 years of professional experience in architecture and service orchestration in critical production environments. I build resilient, fault-tolerant microservices using Python and Kotlin/JVM stacks—especially Spring Boot and FastAPI—and I’m comfortable taking systems from design through operational hardening.

Recently, I’ve been focused on LLM agent infrastructure, including MCP servers and multi-agent orchestration, translating tool needs into reliable interfaces that production systems can safely execute. I’ve proven I can automate manual work at scale—building high-availability pipelines and a provisioning microservice that reduced manual configuration time by over 95%—while improving reliability with event-driven patterns, circuit breakers, retries, and observability.

Experience

Work history, roles, and key accomplishments

Current

Backend Engineer

Current

Sparklight

Oct 2024 - Present (1 year 8 months)

Built and operationalized a monorepo of Python MCP servers (FastMCP over STDIO) integrating enterprise network and ITSM systems as tool-callable interfaces for LLM agents, including the MCP layer powering Sparknet for production network actions. Developed the Calix Provisioning Platform (Python/FastAPI, Docker, Ansible AWX), automating broadband device provisioning and cutting manual configuration

Python fastAPI Kafka Microservices Docker

Prompt Engineer (LLM Trainer)

Scale AI

Apr 2024 - Sep 2025 (1 year 5 months)

Performed RLHF for LLMs on software-development tasks, evaluating and ranking model outputs for code development, bug detection, and refactoring across Python, Go, JavaScript, and React.

Reinforcement Learning LLM Evaluation Prompt Engineering Refactoring Python Go JavaScript

Backend Software Developer

Quinto Andar

Apr 2024 - Oct 2024 (6 months)

Engineered an event-driven PP Multi classification pipeline in Kotlin/Spring Boot/Kafka that replaced manual eligibility checks, processing ~3,000 property listings/day and automating decisions for ~1,000 owners/month via a chain-of-responsibility and PostgreSQL outbox pattern. Designed resilience with Resilience4j (circuit breakers, rate limiting, retries), Kafka retry handling, and DLQ reprocess

Kotlin Spring Boot Kafka Resilience4j Outbox Pattern PostgreSQL Event Driven Architecture Circuit Breakers

Software Developer

Nelogica

May 2020 - Feb 2024 (3 years 9 months)

Designed and deployed a Delphi-based WebSocket messaging service that decoupled the proxy layer from backend systems, reducing memory usage on proxy infrastructure by 23%. Refactored legacy Delphi code and redesigned Firebase-powered mobile notification data structures to cut memory usage by 71%, and rebuilt an instrumentation monitoring system for 60+ services and 1,000+ servers to reduce provisi