Skip to content

README.md

Latest commit

 

History

History
182 lines (135 loc) · 7.22 KB

README.md

File metadata and controls

182 lines (135 loc) · 7.22 KB
Raw

ML Engineer Learning Track

"Build the systems that make machine learning work in the real world."

ML Engineers build, deploy, and maintain production-grade ML systems at scale. They bridge the gap between data science research and real-world software products. This is one of the highest-paying engineering roles in tech.

Track Overview

PHASE 1: Software Engineering Foundations (2-3 months)
    Python SWE → Git → Docker → System Design
         ↓
PHASE 2: ML Fundamentals (3-4 months)
    ML Algorithms → Deep Learning → Model Evaluation → PyTorch
         ↓
PHASE 3: MLOps & Production (4-6 months)
    Pipelines → Model Serving → Monitoring → Cloud ML
         ↓
PHASE 4: Advanced (3-5 months)
    Distributed Training → LLM Fine-tuning → Platform Engineering

Prerequisites

  • Solid Python programming (OOP, decorators, testing, type hints)
  • Basic Linux/terminal comfort
  • Understanding of REST APIs
  • Basic Git workflow (commit, branch, merge)
  • Basic ML concepts (what a model is, train/test split, etc.)

Don't have these? Start at Foundations →

Skills You Will Build

Technical Skills

Category Skills
Programming Python (expert), Bash, SQL, Scala (optional)
ML Frameworks PyTorch, TensorFlow, scikit-learn, XGBoost
MLOps MLflow, Weights & Biases, DVC, Kubeflow
Model Serving FastAPI, BentoML, TorchServe, Triton
Infrastructure Docker, Kubernetes, Terraform
Data Engineering Apache Airflow, Spark, Kafka, dbt
Cloud AWS SageMaker / GCP Vertex AI / Azure ML
Monitoring Evidently AI, Prometheus, Grafana
Feature Stores Feast, Tecton
Testing pytest, Great Expectations, model testing

Beginner Phase — Software Engineering for ML

Goal: Write production-quality Python, understand containers, and build simple ML APIs.

Duration: 2-3 months (10-15 hrs/week)

Week Topic Resource Project
1-2 Advanced Python (OOP, testing, packaging) Python SWE Guide Refactor a messy script
3-4 Git for teams Git Deep Dive Collaborate on a shared repo
5-6 Docker fundamentals Docker Guide Containerize a Python app
7-8 REST APIs with FastAPI FastAPI Guide Build a prediction API
9-10 ML model basics (sklearn + PyTorch) ML Refresher Train + serialize a model
11-12 End-to-end: Train + Serve First ML API Project Full mini-project

→ Start Beginner Phase

Intermediate Phase — MLOps Core

Goal: Track experiments, build ML pipelines, and deploy models to production.

Duration: 3-4 months (10-15 hrs/week)

Week Topic Resource Project
1-2 Experiment Tracking (MLflow) MLflow Guide Track 10 experiments
3-4 Data Version Control (DVC) DVC Guide Version a dataset + model
5-6 Pipeline Orchestration (Airflow) Airflow Guide Automated retraining pipeline
7-8 Model Serving at Scale Serving Guide Deploy model to prod
9-10 Model Monitoring Monitoring Guide Detect data drift
11-12 Feature Stores (Feast) Feature Store Guide Build feature pipeline
13-14 CI/CD for ML ML CI/CD Guide GitHub Actions for ML

→ Start Intermediate Phase

Advanced Phase — Production Scale & LLM Systems

Goal: Design ML platforms, train at scale, fine-tune LLMs, and handle enterprise ML challenges.

Duration: 4-6 months (10-15 hrs/week)

Week Topic Resource Project
1-3 Kubernetes for ML K8s Guide Deploy ML on K8s
4-6 Distributed Training Distributed Training Multi-GPU training job
7-9 LLM Fine-tuning (LoRA/QLoRA) LLM Fine-tuning Guide Fine-tune Llama on custom data
10-12 ML Platform Architecture Platform Architecture Design ML platform spec
13-15 Inference Optimization Optimization Guide Reduce latency by 50%

→ Start Advanced Phase

Projects

Beginner Projects

  • ML Model as REST API (FastAPI + Docker)
  • Model Comparison Dashboard (MLflow + 5 models)
  • Automated Data Quality Checker

Intermediate Projects

  • End-to-End ML Pipeline (Airflow + MLflow + FastAPI)
  • Model Monitoring System with Drift Detection
  • A/B Testing Framework for ML Models

Advanced Projects

  • LLM Fine-tuning Pipeline (Llama/Mistral on custom data)
  • Real-time Feature Store with Kafka + Feast
  • ML Platform Design Document + Proof of Concept
  • Multi-model Serving System with Kubernetes

→ See All Projects

Skills Checklist

Beginner Level

  • Can write testable, modular Python code
  • Understand Docker: build images, run containers, docker-compose
  • Can build a REST API with FastAPI
  • Can serialize and load ML models (pickle, joblib, ONNX)
  • Know Git workflows: branching, PRs, code review

Intermediate Level

  • Track ML experiments with MLflow
  • Version datasets and models with DVC
  • Build Airflow DAGs for ML workflows
  • Deploy a model API with Docker + cloud
  • Detect data drift and model degradation
  • Build a basic feature pipeline

Advanced Level

  • Deploy ML workloads on Kubernetes
  • Run distributed training across multiple GPUs
  • Fine-tune an LLM using LoRA
  • Design an end-to-end ML platform architecture
  • Optimize model inference latency (quantization, batching, caching)

Recommended Resources

Free Books & Docs

Resource Notes
Full Stack Deep Learning Best-in-class MLOps curriculum
Made With ML End-to-end production ML (GitHub: GokuMohandas/Made-With-ML)
MLOps Zoomcamp Free 9-week MLOps course
Designing ML Systems Chip Huyen's seminal book

Certifications Worth Getting

  • AWS Certified Machine Learning Engineer — Associate (newer, hands-on SageMaker/MLOps focus)
  • AWS Certified Machine Learning — Specialty
  • Google Professional Machine Learning Engineer
  • Microsoft Certified: Azure Data Scientist Associate
  • Databricks Certified Machine Learning Associate / Professional
  • NVIDIA DLI certificates (deep learning, LLM inference, MLOps on GPU infra)

Back to: Main README | Role Comparison