Hello world!
My name is Matheus Bortoletto. Most people online know me as darkm. I'm a software engineer from Brazil, and this is where I'll be sharing what I learn, what I build, and what I think about when designing systems that need to scale.
What I do
I sit at the intersection of code and strategy. My day-to-day isn't just writing features. It's deciding what to build, how it should fit together, and making sure the technical direction actually serves the business.
That means I spend a lot of time on:
- System architecture: choosing the right boundaries, communication patterns, and trade-offs for systems that need to hold up under real-world pressure.
- Database optimization: query tuning, indexing strategies, schema design, knowing when to denormalize and when to hold the line.
- AWS: networking, IAM, cost optimization, multi-region strategies. I've seen what works at scale and what turns into an invoice nightmare.
- Observability: structured logging, distributed tracing, metrics pipelines. If you can't see what's happening, you can't fix it.
- Event-driven systems: message queues, Temporal workflows, building systems that are resilient to failure and don't lose data.
- Networking and security: VPNs, private connectivity, zero-trust patterns. Keeping things locked down without making life miserable for the team.
My stack
Go, TypeScript, React, Terraform, PostgreSQL, Redis, Kafka, Temporal, Docker, Kubernetes, Datadog, Grafana, AWS.
Agentic software
This is where I'm spending a lot of my energy right now. AI agents are moving fast, and the infrastructure to run them reliably is still being figured out. I'm deep into building agentic systems that actually work in production, not just demos.
That means thinking about:
- Orchestration patterns for multi-agent workflows
- Tool use, memory, and context management at scale
- Guardrails that keep agents useful without letting them go off the rails
- Evaluation and observability for non-deterministic systems
- Cost management when every call burns tokens
The engineering challenges here are genuinely interesting. It's not just prompt engineering. It's systems design, with all the usual concerns around reliability, latency, and failure modes, plus a whole new set of problems that didn't exist two years ago.
I'll be writing a lot about this.
What's next
I have a backlog of topics I want to write about, but I'm not going to promise a schedule. I'll write when I have something worth saying.
If you want to follow along, you're already in the right place.
Welcome to darkm.dev. Let's build something.