prompt
FreeProduction ML system design prompt for AI assistants
FreeFree tier
About prompt
This is a comprehensive system prompt designed for AI assistants to act as expert ML systems architects. It outlines a structured analysis process covering problem definition, data pipeline architecture, model training strategy, inference deployment, monitoring, and cost optimization. The prompt emphasizes production-grade considerations for classical ML, deep learning, LLMs, and ensemble methods, including trade-offs between accuracy, latency, and cost. It is part of the ai-boost/awesome-prompts open-source collection on GitHub.
Key Features
ML system design and architecture (data pipelines, training, inference, monitoring)
Model selection and evaluation (classical ML, deep learning, LLMs, ensemble methods)
Feature engineering and feature stores
Data quality and labeling strategies
Model training infrastructure (distributed training, hyperparameter optimization)
Inference optimization (latency, throughput, cost)
MLOps and model deployment (versioning, A/B testing, rollback)
Monitoring and observability (model drift, data drift)
LLM fine-tuning and adaptation
Cost optimization and resource allocation
Pros & Cons
Pros
- Covers the full ML lifecycle from problem definition to monitoring
- Includes practical considerations like cost, latency, and resource constraints
- Structured analysis process ensures thoroughness and reproducibility
- Adaptable across classical ML, deep learning, LLMs, and ensemble methods
- Addresses both technical and operational aspects (MLOps, observability)
Cons
- Assumes significant prior knowledge of ML systems and infrastructure
- May be too detailed or overwhelming for simple or prototype projects
- Not tailored to any specific framework or cloud provider
Best For
Designing production machine learning infrastructure from scratchSelecting and evaluating models for specific business problemsOptimizing inference pipelines for latency and costSetting up MLOps workflows including versioning, A/B testing, and rollbackBuilding robust data pipelines with quality checks and feature storesFine-tuning and deploying large language models