Diploma in LLM Systems & RAG Engineering
Production retrieval-augmented generation, multi-agent orchestration, evaluation harnesses, and enterprise LLM infrastructure.
About this program
Master the engineering practices required to build, evaluate, and scale production LLM systems. Covers hybrid retrieval, reranking, chunking strategies, semantic caches, multi-agent graphs, and deterministic evaluation gates.
Student ratings
Outstanding — 96 verified Canadian graduates rated this program 5.0/5. Reviews emphasize the applied capstone, instructor responsiveness, and career outcomes.
- 5★94
- 4★0
- 3★2
- 2★0
- 1★0
Who this program is for
- →Practitioners already shipping generative-ai work who want depth
- →Senior engineers, data scientists, and technical leads
- →Canadian residents seeking a verifiable diploma credential
Topics you'll cover
6 modules across 6 months — 29 lessons in total.
Full syllabus
Module 1 · Module 1: Advanced Ingestion, Chunking & Embeddings▾
- L1Parsing messy PDFs, tables, slides, and scanned forms with OCR pipelines
- L2Semantic, recursive, agentic, and late-chunking strategies
- L3Embedding model benchmarks: dense vs. sparse vs. multi-vector (ColBERT)
- L4Vector index topologies: HNSW, IVF-PQ, and partition indexing at scale
- L5Module 1 Capstone: Enterprise financial filing ingestion engine with table extraction
Module 2 · Module 2: Hybrid Retrieval & Reranking Engines▾
- L1Reciprocal Rank Fusion (RRF) and hybrid search architectures
- L2Cross-encoder reranking: Cohere, BGE, and custom transformer rerankers
- L3Query expansion, HyDE (Hypothetical Document Embeddings), and sub-query routing
- L4Document-level metadata filtering, access control (RBAC), and multi-tenancy
- L5Module 2 Capstone: Sub-second legal discovery engine over 100,000 regulatory documents
Module 3 · Module 3: Structured Outputs, Function Calling & Guardrails▾
- L1Constrained decoding, grammars, and JSON Schema schema enforcement
- L2Advanced tool calling with validation, schema fallback, and retries
- L3Input/output guardrails: NeMo Guardrails, LlamaGuard, and PII masking
- L4Prompt versioning, regression testing, and semantic caching architectures
- L5Module 3 Capstone: Automated banking transaction reconciliation agent with strict output schemas
Module 4 · Module 4: Multi-Agent Orchestration & State Graphs▾
- L1State machines and cyclic graph execution (LangGraph, AutoGen)
- L2Plan-and-solve vs. ReAct vs. reflexive agent architectures
- L3Long-term memory management: episodic, procedural, and working memory
- L4Human-in-the-loop approval workflows, timeouts, and rollback states
- L5Module 4 Capstone: Multi-agent software code audit and remediation system
Module 5 · Module 5: RAG Evaluation, Observability & Cost Engineering▾
- L1Ragas evaluation triad: Faithfulness, Answer Relevance, Context Precision
- L2Automated synthetic golden test set generation
- L3Full-stack LLM tracing with OpenTelemetry, Langfuse, and Arize Phoenix
- L4Prompt compression, token budget management, and model tier routing
- L5Module 5 Capstone: CI/CD evaluation gate blocking regressions in pull requests
Module 6 · Module 6: Capstone Practicum & Industry Defense▾
- L1System specification defense and threat modeling
- L2Production deployment with load testing and latency SLAs
- L3Verification of security, evaluation matrices, and cost model
- L4Live jury defense with senior Canadian AI engineering leaders
What you'll be able to do
- ●Design and deploy hybrid retrieval pipelines combining BM25, dense vectors, and cross-encoders
- ●Build autonomous multi-agent workflows with state machines, tool calling, and human-in-the-loop gates
- ●Establish automated LLM evaluation harnesses using Ragas, DeepEval, and deterministic synthetic test sets
- ●Implement token-level caching, prompt compression, and aggressive latency/cost optimization
- ●Structure strict JSON schema validation and structured output contracts
- ●Deliver Canadian enterprise-ready RAG with document-level RBAC and tenant isolation
Career paths after graduation
Frequently asked questions
How much does the Diploma in LLM Systems & RAG Engineering cost?▾
Tuition is $2,800 CAD, paid once. You can pay in full at checkout or choose an interest-free monthly plan. A 30-day refund window applies from your enrollment date.
How long is the Diploma in LLM Systems & RAG Engineering program?▾
Self-paced. Most students complete it in 6 months at roughly 7 hours per week, but you can go faster or slower — you keep lifetime access.
What are the prerequisites?▾
Strong Python or TypeScript backend development experience; Familiarity with REST APIs, asynchronous programming, and databases; Basic conceptual understanding of large language models and embeddings
Is the diploma recognized in Canada?▾
Yes. Graduates receive the Altaris AI Academy Diploma in generative-ai — a verifiable credential with a unique certificate number you can publish on LinkedIn and that any employer can verify at altarisai.org/verify.
What is the refund policy?▾
Full refund within 30 days of enrollment, no questions asked. After day 30, prorated refunds are available per our Refund Policy.
Who teaches the program?▾
Working Canadian AI practitioners — not academics. Every module is built and reviewed by a lead instructor working in the field today.