Senior Backend Engineer | Distributed Systems & Data Infrastructure

I engineer high-concurrency distributed systems and fault-tolerant data pipelines. I specialize in scaling Python/Go backends, ruthlessly optimizing cloud infrastructure footprint, and integrating applied ML to drive system efficiency at scale.

• Leadership Profile: Fast-tracked to Senior Engineer (16 months) for stabilizing and rebuilding a 5-person engineering pod during a period of 60% turnover.
• Operational Scale: Accountable for production systems processing 3B+ monthly events and 120M+ daily requests with 99.9% availability.
• Core Arsenal: Python (Expert), Go, PostgreSQL, AWS (ECS, Lambda, RDS), Terraform.
• Location: Recife, Brazil (Remote).

• Zero-Downtime Migrations: Architected critical infrastructure migrations, including serverless Lambda to ECS containers (handling 120M+ req/day) and PostgreSQL 15.12→18.1 (2M+ records/day), maintaining complete system availability.
• Cost Engineering at Scale: Slashed AWS compute spend by 50% through strategic Lambda→ECS containerization and reduced RDS operational costs by 25% via version optimization and right-sizing.
• ML-Optimized Pipelines: Engineered an SGD Regressor for predictive payload prioritization (20% extraction increase) and deployed a Reinforcement Learning Multi-Armed Bandit (35% scraping efficiency gain).

The Challenge: Backend demonstrating production patterns for AI agent observability, turning opaque execution pipelines into queryable, evaluated traces with strict data integrity.

• Tech: Python 3.12+, Django, Django REST Framework, PostgreSQL 18, Huey 2.6, Docker, Prometheus, mypy (strict mode)
• Impact: Idempotent ingestion, immutable run semantics, durable async failures. These observability patterns are designed to withstand retry storms or partial failures.

The Challenge: Demonstrating production patterns for high-concurrency HTTP workloads—circuit breaker protection, backpressure handling, and graceful shutdown—to solve the unbounded gather() anti-pattern that crashes systems under load.

• Tech: Python 3.12+, aiohttp, asyncio, pytest, Poetry, mypy (strict mode), GitHub Actions
• Impact: 20x throughput gain (130 → 2,500 RPS) with cascade failure prevention via circuit breaker state machine, bounded concurrency enforcement, and graceful degradation under load.

The Challenge: Web-scale data extraction that maintains constant memory footprint regardless of crawl size, solving the classic scraper OOM failure mode.

• Tech: Go, Colly, Worker Pools, Prometheus, LRU Deduplication, Buffered Channels
• Impact: 2.2M items/sec throughput with constant memory footprint via intentional backpressure design

The Challenge: Empirical proof of the most efficient PostgreSQL bulk-read strategies, replacing anecdote with benchmarked evidence for architectural decisions.

• Tech: Python, asyncpg, multiprocessing, Pydantic, PostgreSQL 16
• Impact: 124K rows/sec (5.3x baseline) via multiprocessing, or 97% memory reduction via async streaming

• Anthropic: MCP: Advanced Topics & Claude Code in Action (2026)

I am always interested in discussing distributed systems architecture, zero-downtime migrations, ML-driven optimization, or high-scale data infrastructure. If you are solving complex problems in these domains, let's connect.

• Email: alumlira@gmail.com
• LinkedIn: linkedin.com/in/aluiziolira