AI

ESIA

School of Artificial Intelligence

Worldwide cohort
Students across time zones
HomeMaster’sTrainingsProjectsResearchBlogAboutContact

Master’s Program

Applied ML Engineer

Modeling + engineering to ship reliable ML features

Overview

This track is designed for people who want to build ML features that survive in real systems. You focus on practical ML, evaluation, inference constraints, and engineering patterns used to deliver stable model services.

Key outcomes

  • Build robust ML baselines and iterate fast with clean experiments
  • Design inference pipelines with performance & reliability in mind
  • Deploy services and manage versioning correctly
  • Write maintainable ML code with tests and reusable modules

Format

  • Hands-on labs + deliverables
  • Remote-friendly (worldwide cohorts)

Tools

  • Python, scikit-learn
  • FastAPI
  • Docker
  • Git/GitHub
  • MLflow basics

Detailed program

Engineering-oriented ML: build reliable model features with strong evaluation, clean architecture and deployment constraints in mind.

Module 1 — ML Baselines & Evaluation Discipline

Build strong baselines and evaluate correctly before optimizing.

2 weeks

What you’ll learn

  • Pipeline setup: preprocessing, splits, cross-validation
  • Metrics selection by use case + thresholding basics
  • Error analysis: slices, robustness checks, leakage detection
  • Experiment discipline: clean comparisons and tracking habits

Skills you’ll gain

Baseline buildingEvaluation disciplineError analysisExperiment rigor

Module 2 — Training/Inference Architecture

Design code that works both in training and production inference.

2–3 weeks

What you’ll learn

  • Train/inference parity and feature pipelines
  • Reusable transforms and data contracts
  • Model packaging and input schemas
  • Performance basics: batch inference mindset

Skills you’ll gain

ML architectureData contractsModel packagingInference constraints awareness

Module 3 — Deployment Patterns (Service-first)

Ship ML as an API with stability and maintainability.

2–3 weeks

What you’ll learn

  • FastAPI endpoints + validation + version route
  • Dockerizing the service for consistent runtime
  • Logging and simple observability practices
  • Rollback mindset and safe iteration

Skills you’ll gain

API deploymentDocker workflowService reliability mindsetRelease/rollback thinking

Module 4 — Production Quality (Testing & Maintainability)

Make your ML codebase maintainable like real engineering teams.

1–2 weeks

What you’ll learn

  • Unit tests for preprocessing and business rules
  • Basic integration tests for API payloads
  • Refactoring patterns and clean module boundaries
  • Documentation: README, usage, constraints, known limitations

Skills you’ll gain

Testing ML systemsMaintainable codeRefactoring disciplineDocumentation standards

Capstone — ML Feature Service

Deliver a stable ML service ready for integration in real systems.

2–3 weeks

What you’ll learn

  • Build an end-to-end ML pipeline + API + Docker image
  • Add tests and a minimal CI workflow
  • Document integration usage and model behavior
  • Demo + portfolio packaging

Skills you’ll gain

End-to-end ML deliveryService integration readinessTesting + CI basicsPortfolio-grade demo