Machine Learning Engineer

Evlo AI; Atlanta (GA); In loco; USD 150.000 - 190.000; Tempo pieno

Descrizione del lavoro

Evlo AI in Atlanta is seeking a machine learning engineer to own production-ready models from training through deployment and monitoring, collaborating with data scientists and platform teams.

You will design, train, and evaluate models for classification, ranking, NLP and recommendations, build data pipelines, deploy on SageMaker/Vertex AI or Kubernetes, and implement monitoring and automated retraining.

Competenze

• 3–6 years of production ML engineering experience.
• Proficient in Python with testing and code reviews.
• Deep proficiency with PyTorch or TensorFlow; familiarity with scikit-learn.
• Production experience with cloud ML platforms: AWS SageMaker, Vertex AI, or Azure ML.
• Experience with Docker/Kubernetes and orchestration tools.
• Familiarity with MLflow or similar for experiment tracking.

Mansioni

• Design, train, and evaluate production ML models with clear ownership of accuracy, latency, and cost.
• Build and maintain data pipelines for feature engineering and training datasets.
• Deploy models end-to-end using SageMaker, Vertex AI, or Kubernetes-based stacks with canary/shadow deployments.
• Instrument model monitoring for drift and performance, with automated alerts and retraining triggers.
• Optimize inference: batching, quantization, distillation, hardware-aware serving.
• Collaborate with data scientists and MLOps engineers to establish experiment tracking and reproducible workflows.
• Contribute to code reviews and ML engineering best practices across the team.

Conoscenze

Python engineering

PyTorch/TensorFlow

Data pipelines

Model evaluation

Experiment tracking

Latency optimization

Collaboration with ML team

Strumenti

Docker

Kubernetes

Airflow

Kubeflow

Dagster

MLflow

Descrizione del lavoro

About The Role

The role sits at the intersection of model development and production systems: taking ML workloads from prototype to monitored, reliable services running at scale.

The team ships models that support real product decisions daily, and this role will own critical pieces of that pipeline - from training and evaluation through deployment, serving, and continuous monitoring - working alongside data scientists, backend engineers, and platform teams.

Key Responsibilities

• Design, train, and evaluate production ML models - classification, ranking, NLP, and recommendation use cases - with clear ownership of accuracy, latency, and cost tradeoffs


• Build and maintain data pipelines in Python and PySpark/Spark for feature engineering, training datasets, and online/offline consistency


• Deploy models end-to-end using AWS SageMaker, Vertex AI, or Kubernetes-based serving stacks, including canary releases, shadow deployments, and rollback procedures


• Instrument model monitoring for data drift, feature skew, and performance regression; build automated alerting and retraining triggers


• Optimize inference performance: batching, quantization, distillation, and hardware-aware serving for latency-sensitive workloads


• Collaborate with data scientists and MLOps engineers to establish experiment tracking, model versioning (MLflow or equivalent), and reproducible training workflows


• Contribute to code reviews, architecture discussions, and ML engineering best practices across the team

What We Are Looking For

• 3–6 years of experience in machine learning engineering or applied ML, including multiple production model deployments


• Strong Python engineering skills; comfort with testing, code review, and software engineering fundamentals beyond notebooks


• Deep proficiency with PyTorch or TensorFlow, plus hands-on experience with scikit-learn and classical ML methods


• Production experience with at least one major cloud ML platform: AWS SageMaker, GCP Vertex AI, or Azure ML


• Solid grasp of ML fundamentals: evaluation design, regularization, bias-variance tradeoff, and distribution shift


• Experience with containerized ML services (Docker, Kubernetes) and workflow orchestration (Airflow, Kubeflow, or Dagster)


• Bonus: experience with LLM fine-tuning or serving (vLLM, Triton), real-time feature stores (Feast, Tecton), or a Master's/PhD in a quantitative field