About

I'm Taynan Rezende, a software engineer with 10+ years building backend systems at scale. My work lives at the intersection of distributed systems design, cloud infrastructure efficiency, and operational observability — the domains where software engineering decisions have compounding, measurable effects on the organizations and users that depend on them.

I've held senior and staff roles at iFood (Latin America's largest food-delivery platform), Coinbase (NYSE-listed US crypto exchange), OOBJ (acquired by Avalara), and TOTVS (Brazil's largest enterprise software company), across engineering challenges that range from high-throughput event pipelines and financial-grade transaction consistency to platform-wide cost optimization and legacy infrastructure modernization.

What I work on — and why it matters at scale

The technical domains I focus on are not narrow specializations — they are foundational infrastructure concerns that affect the reliability, security, and economic efficiency of digital systems across every sector of the US economy.

• Distributed systems resilience — designing for partial failure, idempotency, and graceful degradation under real traffic conditions. Reliable infrastructure is a national security concern: CISA's Critical Infrastructure Security and Resilience framework explicitly identifies operational continuity of digital systems as a priority for the US economy and public safety.
• Cloud cost efficiency — data-model-driven optimization and right-sizing that reduces infrastructure spend by orders of magnitude without sacrificing reliability. The GAO and OMB have repeatedly identified cloud overprovisioning as a major source of waste in both federal IT and commercial enterprise infrastructure — waste that engineering excellence can eliminate.
• Operational observability — structured logging, distributed tracing, and metric-driven alerting that make production systems debuggable under pressure. Executive Order 14028 (“Improving the Nation's Cybersecurity”) mandates enhanced observability and logging standards across US government systems and their software supply chain, and these practices are now baseline requirements for any engineering organization serving regulated industries.
• AI-driven operational efficiency — applied machine learning for automating enterprise workflows and scaling support operations. Executive Order 14110 (“Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence”) positions AI adoption as a strategic national priority; early applied work in this space — like the enterprise AI chatbot I built in 2017, before the current wave of LLM adoption — demonstrates the kind of ahead-of-curve execution that drives real competitive advantage.

Track record

The cases below are selected highlights from a broader body of work spanning backend architecture, platform engineering, data infrastructure, and cloud cost optimization across four companies and ten years. Each was chosen because it is documented, metric-driven, and representative of a repeatable pattern — not because it is the only example of its kind.

• Coinbase · Consulting Engineer · 2025–2026. Designed, built, and optimized backend services in Go for CoinBiz — Coinbase's business onboarding platform — processing KYC/KYB document validation workflows (driver's licenses, government IDs, W-9, business filings) at scale under US financial compliance requirements (FinCEN / BSA). Built event-driven pipelines on AWS Lambda, DynamoDB, and Kafka; hardened secure document storage; and developed Salesforce integrations and operational dashboards tracking onboarding volume, success rates, and compliance signals across high-throughput business customer flows.
• iFood · Staff Software Engineer · 2020–present. Redesigned the DynamoDB data model for an auction event pipeline processing 45 billion writes per month, reducing infrastructure cost from $20,000 to $1,800/month — a 91% reduction ($218,400/year saved) with no infrastructure migration. Full case →
• iFood · Staff Software Engineer · 2020–present. Optimized a campaign messaging pipeline on AWS SQS FIFO in Go, increasing throughput from 7.7 messages/second to 462 messages/second — a 60× improvement — through deterministic deduplication and batching. Full case →
• OOBJ (acq. by Avalara) · Software Engineer · 2018–2020. Led the migration of a decade of Subversion history to Git without losing a single commit, then rebuilt the Jenkins pipeline from the ground up — cutting customer release time from more than one full working day to a few hours, freeing ~400–600 engineer-hours per year previously consumed by release babysitting. Full case →
• TOTVS · Software Engineer · 2016–2018. Built a multi-channel enterprise AI chatbot that took first place in an internal hackathon — before LLMs made this approach mainstream — automating L1 support across Skype, Facebook Messenger, and WhatsApp with intent classification, confidence-based escalation, and knowledge base integration. Full case →

These are the cases with numbers attached. Behind them sit a decade of architecture reviews, platform migrations, reliability improvements, and team-level technical decisions that don't always produce a single quotable metric — but collectively define the kind of engineer who shows up at the hardest problems first.

How I think about the work

The best engineers make systems simpler over time, not more clever. My bias is toward writing less code, shipping smaller changes, and paying attention to the failure modes everyone else is ignoring. The highest-leverage moves I've seen in a decade of production work aren't algorithmic breakthroughs — they're data model changes, batching strategies, and the decision to instrument a system so you can actually see what it's doing at 3am.

I write about these patterns publicly because the gap between what practitioners know and what gets documented is where a lot of avoidable engineering waste lives. The writing on this site is an attempt to close that gap — case by case, with concrete numbers and reproducible techniques.

Elsewhere

You can find me on LinkedIn. For direct messages, the contact page is the fastest way to reach me.