AI/ML systems¶

The production-ML side of the AI stack: training infrastructure, model deployment, MLOps, feature stores, and the cloud platforms that orchestrate it. Where LLMs and GenAI covers language-model specifics, this page covers the broader systems engineering: classical ML, deep learning, GPU infrastructure, ML pipelines, and the certs that test them.

flowchart LR
  D[Data + features] --> TR[Training infra<br/>GPUs, distributed training]
  TR --> REG[Model registry]
  REG --> SRV[Serving:<br/>online / batch / edge]
  SRV --> MON[Monitoring + drift]
  MON -. retrain .-> TR

Learn¶

• LLM basics - the foundation under modern AI systems
• Inference servers - vLLM, TGI, SGLang, llama.cpp, TensorRT-LLM
• Quantization and distillation - shrinking models for cheaper serving
• Embeddings and vector search - retrieval as a system
• Evals for LLMs - the regression test for ML systems
• Multimodal models - vision, audio, video pipelines

Compare¶

• AI/ML services (cloud-native) - SageMaker vs Vertex AI vs Azure ML across training, deployment, monitoring
• GenAI platforms - hosted LLM APIs (Anthropic, OpenAI, Bedrock, Azure OpenAI, Vertex)
• Vector databases - retrieval substrate
• LLM observability - LangSmith, Langfuse, Helicone, Phoenix, Braintrust

Reference¶

• Architecture pattern: AI/ML pipeline - production-shaped ML pipeline from data to serving
• Architecture pattern: data pipeline / ETL - the data layer ML systems depend on

Build¶

• Deploy an ML model - train, register, deploy, monitor
• Run Llama on a single GPU - vLLM, OpenAI-compatible endpoint
• Fine-tune with LoRA - LoRA on a small open model, eval against base
• Set up an eval harness - golden set, regression detection, CI integration

Certify¶

Certs that test ML systems engineering specifically:

Foundational - AWS AI Practitioner - cross-cert GenAI study track - Azure AI Fundamentals (AI-900)

Associate - AWS ML Engineer (MLA-C01) - the production-ML cert - Azure AI Engineer (AI-102) - GCP Machine Learning Engineer - Databricks ML Associate - Databricks-flavored MLOps - Databricks GenAI Engineer Associate - NVIDIA AI Infrastructure & Operations Associate - GPU infra - NVIDIA GenAI/LLM Associate

Specialty / Professional - AWS Machine Learning Specialty (MLS-C01) - Databricks ML Professional - NVIDIA AI Infrastructure Professional - NVIDIA AI Operations Professional - NVIDIA Accelerated Data Science Professional

Roadmap¶

The career-track view: AI/ML Engineer roadmap.

Related topics¶

• LLMs and GenAI - language model specifics
• Databases - feature stores and warehouses
• Observability - model monitoring overlaps with general telemetry