MLOps Engineer

MLOps Engineer — Aether AI Platforms

San Francisco, CA | August 2023 – Present

• Architected and deployed 4 end-to-end Kubeflow pipelines on AWS EKS to automate retraining of recommendation models, reducing retraining cycle latency by 35% and saving 14 engineering hours per week.
• Engineered a centralized feature store using Feast to manage 120+ feature sets across 3 real-time applications, improving inference feature consistency to 99.9% and reducing pipeline maintenance by 28%.
• Configured model monitoring and drift detection using Prometheus and Grafana for 18 production endpoints, decreasing time-to-detect data drift from 4 days to less than 15 minutes.
• Optimized Triton Inference Server deployment configurations, enabling automated batching for 2 core NLP models and cutting average API response latency from 180ms to 42ms under peak loads.

Machine Learning Operations Specialist — Vertex Data Systems

Sunnyvale, CA | June 2021 – July 2023

• Implemented automated CI/CD pipelines using GitHub Actions to package 15+ machine learning models into Docker containers, reducing software deployment failure rates by 34% over a 12-month period.
• Managed model registry using MLflow to track 250+ active model runs, standardizing deployment metadata and accelerating artifact promotion rates from staging to production by 40%.
• Provisioned cloud infrastructure using Terraform, automating the setup of 5 reproducible GPU-accelerated staging environments and reducing environment setup time by 75%.
• Collaborated with data science teams to optimize PyTorch training jobs on AWS EC2, adjusting memory allocation to reduce cloud computing costs by 22% across 3 large-scale deep learning models.

Machine Learning Pipelines, CI/CD, Containerization, Infrastructure as Code, Model Monitoring, Drift Detection, Feature Stores, Kubeflow, MLflow, Feast, Triton Inference Server, Docker, Kubernetes, AWS (EKS, EC2, S3), Terraform, PyTorch, Python, Prometheus, Grafana, Git

Automated Re-training Engine

Role: Lead MLOps Engineer

Tools: Kubeflow, MLflow, AWS EKS, S3, Slack

Designed and deployed a continuous learning pipeline that reduced model drift degradation by 40% and saved 12 hours of manual engineer intervention per model update.

High-Throughput Model Serving Platform

Role: Infrastructure Engineer

Tools: Triton Inference Server, Kubernetes, Terraform, Prometheus

Architected a containerized serving infrastructure that reduced GPU utilization cost by 26% and served over 1.5M API requests daily with 99.9% uptime.

• AWS Certified Machine Learning – Specialty (2023)
• Certified Kubernetes Administrator (CKA) (2022)
• HashiCorp Certified: Terraform Associate (2022)

• Languages: English (Native), Spanish (Elementary)
• Volunteer Work: MLOps advisor for AI for Good non-profit organization (2023-present)
• Availability: 3 weeks notice