TECHNICAL LEARNING GUIDES
AI Engineering Learning Guides
In-depth guides on LLMs, LLMOps, MLOps and production AI operations.
LLM Foundations
Architecture, training, fine-tuning, alignment and model selection.
AI Infrastructure Planning: Compute, Kubernetes, GPUs and Capacity
Plan production AI infrastructure across CPUs, GPUs, Kubernetes, concurrency, autoscaling, availability, capacity and unit economics for ML, LLM, RAG and agent workloads.
Fine-Tuning Large Language Models: SFT, LoRA and QLoRA
Learn how to fine-tune large language models using SFT, LoRA and QLoRA, prepare data, compare baselines and evaluate model behaviour.
Large Language Model Alignment and Evaluation: DPO, RLHF and Release Decisions
Understand DPO, RLHF, preference data and large language model evaluation. Build evidence for a release, revise or reject decision.
Large Language Model Architecture and Training: Tokens, Attention and Pretraining
Learn how large language model architecture works through tokenization, embeddings, causal attention, decoder blocks, pretraining and context trade-offs.
Large Language Model Observability: Traces, Evaluation and Drift
Learn large language model observability through traces, system and quality metrics, production evaluation, drift, SLOs and alerts.
Large Language Model Serving and Inference: Latency, Throughput and Scaling
Understand large language model serving, inference metrics, batching, KV cache, quantization, capacity planning and production scaling.
Large Language Models: How They Work, Adapt and Move Into Production
Understand large language models from tokens and Transformers through fine-tuning, alignment, evaluation and the handoff to production LLMOps.
LLMOps Release Engineering: Versioning, Evaluation Gates and Rollback
Learn LLMOps release engineering for models, adapters and prompts through versioning, evaluation gates, progressive delivery, lineage and rollback.
What Is LLMOps? Large Language Model Operations Explained
Learn what LLMOps means, what its lifecycle manages, how it differs from MLOps and how teams operate large language model systems.
Open-Source Large Language Models: How to Compare and Choose
Compare open-source and open-weight large language models by licence, architecture, hardware, adaptation options and evaluation evidence.
Production AI Operations
Learn how teams release, monitor, recover and govern ML, LLM, RAG and agent systems in production. Covers the 7-stage lifecycle, ownership and sub-discipline boundaries.
Production Large Language Model Reliability: Security, Cost and Incident Response
Design reliable production large language model systems with security controls, cost measurement, capacity planning and incident response.
Production AI / AIOps
Lifecycle, readiness, platform engineering, deployment, governance and FinOps.
AI Agent Reliability: State, Tool Calls, Tracing and Recovery
Design reliable production AI agents using durable state, tool validation, retries, human approvals, recursion limits, tracing, evaluation and recovery patterns.
AI Deployment Architecture: Patterns and Decisions
Choose the right AI deployment pattern — batch, real-time, streaming, asynchronous or long-running agent. Compare latency, freshness, throughput and failure trade-offs with a deployment selector.
Operational AI Governance: Access, Audit, Risk and Runtime Controls
Design operational AI governance using lineage, approvals, access controls, runtime policies, audit evidence and production kill switches across AI systems.
AI Observability and Incident Response Across Models, RAG and Agents
Connect infrastructure, data, model, retrieval and agent signals to AI incident detection, triage, containment, recovery and blameless post-incident review.
AI Platform Engineering: Architecture, Golden Paths and Team Boundaries
AI platform engineering applies platform engineering to AI workloads. Learn the 12-layer platform architecture, golden paths, self-service capabilities and platform-team vs product-team responsibilities.
AI Production Readiness: The Go/No-Go Decision Framework
Decide whether your AI system is production ready. Use a 10-dimension readiness scorecard with critical blockers, SLO formulas and evidence requirements for a go, conditional-go or no-go.
AI Reliability Engineering: SLOs, Failure Modes and Recovery
Define AI reliability with SLOs, error budgets, failure classification and resilience patterns across probabilistic production AI systems. Includes burn-rate formulas.
Build vs Buy an AI Platform: Managed, Open-Source or Hybrid
Compare managed, open-source and hybrid AI platform strategies by maturity, control, TCO, vendor lock-in and operational responsibility. Includes a weighted decision model.
FinOps for AI: Cost Optimization and Unit Economics
Apply FinOps to AI: allocate spend by team and outcome, forecast demand, optimize GPU and token costs, and connect cost to quality and business value.
Model Drift and Automated Retraining Without Unsafe Releases
Detect model drift, distinguish covariate, feature, prediction and concept drift, decide when retraining is justified and release updated models through validation and rollback gates.
MLOps CI/CD: Versioning, Model Promotion and Rollback
Build MLOps CI/CD pipelines that version code, data and models, test candidates, promote through environments and roll back safely. Includes current MLflow practices.
Why RAG Systems Fail in Production — and How to Diagnose Them
Diagnose RAG failures across ingestion, chunking, retrieval, reranking, context, generation and evaluation using a trace-to-fix workflow that separates retrieval from generation failure.