Learn Python AI Application Engineer

Model interface and provider abstraction untuk membangun aplikasi AI Python yang tidak terkunci pada satu vendor/model, tetap typed, observable, testable, dan siap production.

Prompting as Protocol Design

Prompting sebagai protocol design: cara mendesain instruksi LLM yang modular, versioned, testable, auditable, aman, dan bisa dipakai di production AI application.

Structured Output, Schema, and Validation

Structured output, schema design, validation, repair loops, and typed contracts for production-grade Python AI applications.

Tool Calling and Function Contracts

Tool calling, function contracts, authorization, idempotency, approval gates, and auditability for production-grade Python AI applications.

Conversation State and Context Management

16 min

Conversation state, context management, memory boundaries, summarization, context packing, and auditability for production-grade Python AI applications.

Async, Streaming, and Backpressure

Async Python, streaming responses, cancellation, timeout, backpressure, queues, and runtime reliability for production-grade AI applications.

Embeddings and Semantic Representation

Embeddings, semantic representation, similarity, vector records, embedding pipelines, quality diagnostics, and production retrieval foundations for Python AI applications.

Document Ingestion and Parsing Pipelines

Production document ingestion and parsing pipelines for AI applications, including source connectors, canonical elements, provenance, metadata, idempotency, quality gates, and regulatory auditability.

Chunking, Indexing, and Knowledge Modeling

24 min

Chunking, indexing, and knowledge modeling for production-grade RAG systems.

Vector Search, Hybrid Search, and Reranking

20 min

Vector search, hybrid retrieval, reranking, filtering, and ranking pipelines for production-grade RAG.

RAG Pipeline Design

End-to-end RAG pipeline design for production AI applications, including query planning, retrieval orchestration, context assembly, answer contracts, citations, refusal, and observability.

RAG Failure Modes and Diagnostics

20 min

Systematic diagnosis of RAG failure modes across ingestion, chunking, indexing, retrieval, reranking, context assembly, generation, citations, and production operations.

RAG for Enterprise Knowledge Systems

Enterprise RAG knowledge systems: tenancy, permissions, metadata, source authority, freshness, lineage, governance, auditability, and knowledge operations.

Agent Mental Model

Agent mental model for production AI applications: perception, planning, tool use, state, memory, policies, autonomy boundaries, and failure control.

Agent Workflow Orchestration

11 min

Agent workflow orchestration with state machines, graph execution, deterministic nodes, model decision nodes, human approval, checkpointing, retries, interrupts, and production tracing.

Tool Registry, MCP, and Integration Contracts

Tool registry, Model Context Protocol, and integration contracts for safe, typed, observable, and permission-aware AI tool use.

Agent Memory and Long-Running Tasks

Agent memory and long-running task engineering: working state, durable memory, checkpoints, resumability, interrupts, approvals, retention, privacy, and recovery.

Multi-Agent Systems and Boundaries

Multi-agent systems and boundaries: when to use multiple agents, coordination patterns, supervisor routing, handoffs, shared state, failure isolation, evaluation, and anti-patterns.

Evaluation Foundations

12 min

Evaluation foundations for production AI applications: eval-first mindset, datasets, rubrics, metrics, regression gates, scenario design, calibration, and release readiness.

RAG and Agent Evaluation

11 min

Practical evaluation for RAG and agent systems: retrieval metrics, groundedness, citation accuracy, tool correctness, trajectory scoring, safety gates, and diagnostic eval reports.

LLM-as-Judge and Human Review

12 min

LLM-as-judge and human review for AI application evaluation: rubric design, calibration, bias control, disagreement handling, adjudication, quality sampling, and review operations.

Testing AI Applications

10 min

Testing AI applications across deterministic code, prompts, structured outputs, providers, RAG, tools, agents, workflows, safety, regression, and CI release gates.

Observability, Tracing, and Debugging

Observability, tracing, and debugging for production AI applications: GenAI telemetry, prompt/model traces, retrieval traces, tool traces, agent traces, metrics, logs, replay, privacy, and incident diagnosis.

Reliability Patterns for AI Systems

Reliability patterns for AI systems: timeout budgets, retries with jitter, fallback, circuit breakers, bulkheads, rate limits, backpressure, idempotency, graceful degradation, chaos testing, and operational runbooks.

Latency, Cost, and Throughput Engineering

Latency, cost, and throughput engineering for production AI applications: token economics, TTFT, streaming, batching, caching, model routing, retrieval budgets, concurrency, queues, and capacity planning.

Security Threat Modeling for LLM Apps

17 min

Security threat modeling for LLM applications: prompt injection, data exfiltration, insecure tool use, excessive agency, insecure output handling, RAG poisoning, supply-chain risk, and defense-in-depth.

Privacy, Governance, and Auditability

Privacy, governance, and auditability for production AI applications: data classification, consent, retention, lineage, DPIA-style review, model/provider governance, audit trails, policy controls, and regulated workflow defensibility.

Deployment Architecture and Runtime Operations

Deployment architecture and runtime operations for Python AI applications: service topology, model gateway, RAG services, workers, queues, Kubernetes, secrets, rollout/rollback, scaling, health checks, SLOs, and operational runbooks.

AI CI/CD and Readiness Gates

11 min

AI CI/CD and readiness gates for production AI systems: prompt/model/index/tool/workflow versioning, eval gates, security gates, cost gates, release trains, canary, shadow, rollback, and production readiness review.

Enterprise Case Management AI Capstone