Alexandre Paiva SRE / DevOps Engineer

ALEXANDRE PAIVA SRE · DEVOPS · PLATFORM

Clarity under pressure.

I follow the alert to the root cause. Evidence over assumptions.

I'm an SRE/DevOps engineer working with cloud infrastructure, Kubernetes, and automation at Elven Works. I help companies keep systems stable, find bottlenecks, and improve production environments — from deployment and monitoring to incident investigation. My goal is to make operations more reliable, efficient, and easier to maintain.

See the casesDownload résumé (PDF)
Production PostgreSQL: CPU above 80%Firing
97%of the CPU limit, with 88% throttling
  1. CPU throttling confirmed in Prometheus. Memory to spare: RAM is not the bottleneck.
  2. EXPLAIN ANALYZE on the hot queries and a read of 122 slow queries from the log.
  3. Root cause: OFFSET pagination read 119,145 rows to return 15.
  4. Fix split in two: code (cursor pagination) and database (per-table autovacuum).
Real incident, with anonymized data. Read the full case.
38/38synthetic alerts successfully delivered in routing tests
871alert rules audited across 6 production environments
249 GBrestored with verified integrity
16documented engineering and operations cases

Tools that serve reliability.

Cloud infrastructure, platforms and observability. Every tool connects to problems I've worked on.

Cloud and infrastructure

  • AWS
  • Terraform
  • Linux
  • Oracle Cloud

Platforms and containers

  • Kubernetes
  • Docker
  • Helm
  • Argo CD

Observability and automation

  • Grafana
  • Prometheus
  • Python
  • GitHub Actions

Featured cases

Every case follows the same shape: the context, what I did, what changed and what I learned, always with the number that was measured.

  1. 01 / SIGNAL

    The alert existed. The message never arrived.

    From broken routes to verified delivery.

    Read the case
    test alerts delivered after rewiring the routing38/38An alert that reaches nobody doesn't exist: monitoring critical services
  2. 02 / CONTEXT

    A critical alert. But were users affected?

    RED dashboards connect technical symptoms with application behavior.

    Read the case
    of 871 rules had a panel; now every alert leads to the right chart4.6%From the alert to the right chart: infrastructure and RED dashboards for applications
  3. 03 / RECOVERY

    A backup is not a guarantee until restore works.

    Measured, verified and documented recovery.

    Read the case
    restored in 2h41 to prove the backup works249 GBA backup only counts if the restore was tested: RPO, RTO and disaster recovery
  4. Diagnose

    The incident that fixed itself: a production database saturated by pagination

    SaaS messaging API platform; the same pattern later showed up on an AI agents platform

    Read the case
    rows read to return 15: the cause of 4 incidents119,145The incident that fixed itself: a production database saturated by pagination
  5. Diagnose

    Half the telemetry vanished for 50 days, and nobody noticed

    SaaS messaging platform

    Read the case
    of half the telemetry lost, until the shard was writing again50 daysHalf the telemetry vanished for 50 days, and nobody noticed
  6. Build

    Commissions can't fail: containers, IaC and blue/green deploys for a variable-pay SaaS

    SaaS for commission and variable-pay management

    Read the case
    promoted as a batch with blue/green, in production4 servicesCommissions can't fail: containers, IaC and blue/green deploys for a variable-pay SaaS
  7. Recover

    War room: the hypervisor disk filled up and took down 14 production nodes

    SaaS omnichannel customer service platform

    Read the case
    of dead space behind the loss of 14 nodes4.8 TBWar room: the hypervisor disk filled up and took down 14 production nodes

All cases

Grouped by the kind of work, from building to protecting. Consulting clients are anonymized and named only by their market segment.

Build

New infrastructure, from scratch or migrated, that the client can run on their own.

Operate

Keep what already exists working every day, and raise the alarm when it doesn't.

Diagnose

Go from symptom to cause, with a number at every step.

Recover

Prove you can come back, and learn when something goes down.

Protect

Shrink the attack surface without breaking production.

Professional experience

Production reliability and product operations: two complementary perspectives.

ELVEN WORKS / SRE · DEVOPS

Production engineering and operations

Observability, Kubernetes, incident investigation and automation with documented outcomes.

Explore technical cases
CAPYVARA / CO-FOUNDER · LEADERSHIP

Technical co-founder at Capyvara

Partner and technical co-founder of a SaaS startup for the public sector: technical leadership of the team, engineering practices and delivery process.

Read the case

What you can expect from me

Reliability is built deliberately. Less uncertainty, greater visibility and safer decisions — especially when something goes wrong.

Clarity before action

I investigate real impact before deciding what needs to change.

Safer changes

I assess risk, validate changes and plan how to roll back if needed.

Visibility into what matters

I build monitoring that helps teams detect failures and understand service health.

Practices that last

I document decisions and automate routines to reduce dependency on manual intervention.

Let's talk about reliability.

Need better visibility, stability or confidence in your systems? Let's talk about your challenges and how I can help.

E-mailcontato.alexandre1504@gmail.com LinkedInin/alexandre-paiva-080181247 GitHubgithub.com/OpsXandao

Résumés by language

PortuguêsPDFEnglishPDFEspañolDOCXItalianoDOCX

Tell me about your challenge

The button opens your email application to send the message.