Machine Learning Ops, dLocal

• Solid experience owning or designing MLOps platforms or ML infrastructure used by multiple teams
• Strong background in distributed systems and data/stream processing (e.g. Spark, Flink, or similar technologies)
• Experience building production‑grade ML pipelines:
• Experiment tracking, reproducible training and model registry
• CI/CD for models and data pipelines
• Online and batch inference at scale
• Familiarity with cloud‑based ML platforms (e.g. Databricks, SageMaker, Vertex AI, or equivalent) and container‑based deployments
• Strong understanding of observability for ML systems:
• Metrics, logs and traces
• Data and model drift, freshness and quality checks
• Ability to communicate clearly with both technical and non‑technical stakeholders, translating infra and AI/ML trade‑offs into business language

Desirable

• As an MLOps, Technical Referent at dLocal, you will be the senior technical reference for how we build, operate and evolve our ML and AI infrastructure
• Your mission is to enable Data Science and AI teams to take models and AI-powered services from idea to production in a reliable, observable and compliant way
• You will own the technical direction of our MLOps stack, introduce AI safely into our engineering workflows, and help the team scale its impact as usage and complexity grow
• A core part of this role is to use agents and AI services to automate as much as possible of what we do in MLOps — from feature store and platform operations to fraud/anomaly workflows and ML cost optimization — working side by side with the AI team
• This is a hands-on architecture and leadership role: you won’t own product models yourself, but you will deeply influence how every model and AI component is trained, deployed, monitored and run in production
• Define and evolve the end-to-end ML platform architecture (data, training, registry, serving, monitoring, governance) used by multiple squads
• Design standard patterns for:
• Reproducible training pipelines and experiment tracking
• Model packaging, versioning and promotion flows (dev → staging → production)
• Online and batch inference, with safe rollout strategies (canary, shadow, rollback)
• Balance reliability, performance and cost for ML workloads, working closely with SRE/Infra and Finance/FinOps
• Act as the go‑to person for complex MLOps questions: how to structure pipelines, choose serving patterns, or design monitoring and rollback
• Review and challenge designs and deployments for new models and data pipelines, ensuring they follow platform standards and non‑functional requirements
• Partner with Fraud, Anomaly and other product squads to ensure:
• Clear SLAs/SLOs for ML components
• Proper logging, metrics and alerts for incidents and regressions
• Contribute to on‑call readiness: playbooks, dashboards, incident reviews and continuous improvement of our operational posture
• Define infrastructure, contracts and guardrails so that we can safely consume agents and AI services built by the AI team, and extend them when needed from MLOps
• Design patterns and tooling so that AI and agents automate as much as possible of what we do in MLOps, for example:
• Feature platform operations (feature store pipelines, backfills, parity checks, DQ/drift monitoring)
• MLOps platform workflows (training/eval pipelines, promotion gates, rollbacks, documentation and runbook generation)
• Operational flows in Fraud / Anomaly (triage of alerts, log/metric analysis, enrichment of incident context)
• Platform FinOps & cost optimization (suggesting right‑sizing, schedule changes, decommissioning opportunities)
• Contribute to evaluation, observability and safety for these AI‑powered automations (e.g. prompts, policies, redaction, auditability), in close collaboration with dedicated AI teams
• Set and maintain technical standards for:
• Model and data access control, PII handling and redaction
• Auditability of model changes, deployments and runtime behavior
• Environment separation and change management for ML/AI workloads
• Work with InfoSec and Architecture to ensure the platform aligns with regulatory and internal requirements while remaining practical for engineers and data scientists
• Mentor MLOps and Data/ML engineers on:
• System design, reliability and observability
• Good practices for CI/CD, testing and rollback in ML systems
• Lead design and architecture reviews, helping teams de‑risk decisions and converge on simple, robust solutions
• Collaborate closely with:
• Data Science squads and the AI Team (to understand needs and shape the platform)
• SRE/Infra (for capacity, reliability, networking and security)
• Product/Engineering leaders (to align roadmap, trade‑offs and priorities)