Serving Stacks
Notes and artifacts will collect practical tradeoffs, measurement patterns, and architecture implications.
Core area
AI Inference
Model serving, latency, throughput, KV-cache pressure, GPU/cloud economics, and AI factory architecture.
Thesis
Inference is where AI product quality, user experience, margin, and infrastructure reality meet. Good systems require measured tradeoffs across latency, throughput, memory, batching, scheduling, hardware, and deployment topology.
Core areas
Latency and Throughput
Notes and artifacts will collect practical tradeoffs, measurement patterns, and architecture implications.
KV Cache and Memory Pressure
Notes and artifacts will collect practical tradeoffs, measurement patterns, and architecture implications.
Batching and Scheduling
Notes and artifacts will collect practical tradeoffs, measurement patterns, and architecture implications.
Cost Models
Notes and artifacts will collect practical tradeoffs, measurement patterns, and architecture implications.
GPU Cloud and AI Factory Notes
Notes and artifacts will collect practical tradeoffs, measurement patterns, and architecture implications.
Benchmarking Methodology
Notes and artifacts will collect practical tradeoffs, measurement patterns, and architecture implications.
Reliability and Observability
Notes and artifacts will collect practical tradeoffs, measurement patterns, and architecture implications.
Projects and artifacts
ai-hub
Practical learning and building hub for modern AI systems, including inference engineering, agents, security, EvalOps, and model architecture notes.
AI systemsInferenceEvalOps
What it demonstrates: How a broad AI systems knowledge base can organize production patterns, labs, and technical reading paths.
ga-gcp-ai-journal
Working journal for learning, building, and documenting generative AI workflows on Google Cloud.
GCPJournalAI workflows
What it demonstrates: How daily technical notes can capture architecture tradeoffs and experiment results without becoming a scratchpad.
ga-gcp-ai-labs
Hands-on Google AI and GCP lab projects, experiments, and reference implementations.
GCPVertex AILabs
What it demonstrates: How cloud AI labs can stay structured around validation, repeatable setup, and deployable patterns.
ga-gcp-ai-patterns
Cloud AI patterns and labs around production deployment.
GCPAI patternsCloud
What it demonstrates: How cloud primitives shape deployable AI systems.
inference-engineering
Inference engineering notes and examples around serving, latency, throughput, and deployment topology.
InferenceServingGPU economics
What it demonstrates: Inference as the point where product quality, user experience, and infrastructure economics meet.
inference-engineering-book
Book workspace for AI Inference Engineering.
BookInferenceArchitecture
What it demonstrates: Long-form systems treatment of model serving and AI factory architecture.