Model Online / On-Site
Duration
Modules 6 modules

Course Objective & Scope

Bridging the gap between a working notebook demo and a reliable, low-latency enterprise AI system requires rigorous engineering practices. This intensive bootcamp takes engineers through commercial APIs (OpenAI, Claude, Gemini) and open-weight models (Llama, Qwen, Mistral), production-grade RAG architectures (semantic chunking, hybrid retrieval, cross-encoder reranking), stateful agent systems with LangGraph and MCP, parameter-efficient fine-tuning (LoRA/QLoRA) and high-throughput serving with vLLM/SGLang.

Through hands-on coding labs and a capstone build, participants implement automated evaluation pipelines, OWASP LLM security guardrails and end-to-end LLMOps observability.

Who Should Attend?

  • Software engineers and AI engineers building production LLM applications
  • Data scientists, ML engineers and backend/platform developers
  • Technical leads and solution architects designing enterprise GenAI systems
  • Prerequisite: Solid Python programming skills and familiarity with REST APIs

Key Takeaways

  • Build resilient LLM service layers with structured outputs, streaming, prompt caching and fallback routing
  • Engineer high-precision RAG pipelines using document parsing, hybrid dense/sparse search, metadata filtering and reranking
  • Develop stateful, tool-calling agentic workflows using LangGraph and Model Context Protocol (MCP)
  • Make evidence-based trade-offs between context engineering, RAG and LoRA/QLoRA fine-tuning
  • Instrument applications with automated evals (RAGAS/DeepEval, LLM-as-a-judge), runtime guardrails and LLMOps tracing

Curriculum

LLM Engineering Foundations, Model Selection and API Architecture
  • Transformer inference mechanics, KV-cache, context window trade-offs and token economics
  • Structured outputs (Pydantic/JSON schema), streaming, parallel tool calling and resilient retries
Production-Grade Advanced RAG Architectures
  • Parsing complex enterprise documents (tables, layouts) and semantic/hierarchical chunking
  • Embedding selection, vector indexing (HNSW), BM25 + dense hybrid retrieval and cross-encoder reranking
  • Query transformation, multi-hop retrieval and citation-grounded generation
Agentic Workflows, LangGraph and MCP Integration
  • State graphs, cyclic execution, short/long-term memory checkpoints and human-in-the-loop interrupts
  • Building and consuming Model Context Protocol (MCP) servers for secure tool execution
Systematic Evaluation, Guardrails and LLM Security
  • Curating golden datasets, retrieval/generation metrics and automated LLM-as-a-judge pipelines
  • OWASP Top 10 for LLM defences: prompt injection mitigation, PII redaction and input/output guardrails
Open-Weight Models, Fine-Tuning (LoRA/QLoRA) and High-Throughput Serving
  • Dataset curation, SFT/preference tuning with QLoRA and adapter management
  • Quantization (AWQ/FP8) and serving models at scale with vLLM and SGLang
LLMOps, Observability and Capstone Build
  • OpenTelemetry tracing (Langfuse/Phoenix), prompt versioning, semantic caching and cost monitoring
  • End-to-end architecture review and live demonstration of the capstone GenAI application

Other Courses in This Category

  • AI-Native Software Engineering Academy

    Transforming the full software development lifecycle (SDLC) into an AI-native discipline: repo context rules, IDE/CLI coding agents, legacy modernisation, automated testing, DevSecOps and team governance.

    Corporate Academies & Programme Models Corporate
    Live Course
    6 modules
    View
  • AI Governance, Risk and Assurance Academy

    Enterprise AI governance and assurance academy aligned with the EU AI Act, ISO/IEC 42001, NIST AI RMF and data privacy laws: AI inventory, risk tiering, technical red teaming, vendor due diligence and operating controls.

    Corporate Academies & Programme Models Corporate
    Live Course
    6 modules
    View
  • Data and AI Platform Engineering Academy

    Enterprise data and AI platform engineering academy: modern lakehouse, AI-ready data governance, vector and context layers, MLOps/LLMOps pipelines, AI gateways, model serving and FinOps.

    Corporate Academies & Programme Models Corporate
    Live Course
    6 modules
    View
  • Human Skills for the AI Era Academy

    Multi-module academy for durable human capabilities in the AI era: problem framing, critical evaluation of AI outputs, ethical judgement, stakeholder storytelling, facilitation and learning agility.

    Corporate Academies & Programme Models Corporate
    Live Course
    6 modules
    View
  • Executive AI Leadership Academy

    Multi-module AI leadership academy for executives and business unit heads: AI strategy, use-case portfolio and unit economics, operating model, EU AI Act/ISO 42001 governance and an enterprise AI roadmap.

    Corporate Academies & Programme Models Corporate
    Live Course
    6 modules
    View
  • AI Fluency and Business Productivity Academy

    Comprehensive AI literacy and productivity academy for business teams: prompt and context engineering, document and data analysis, custom assistants, verification guardrails and a department workflow project.

    Corporate Academies & Programme Models Corporate
    Live Course
    6 modules
    View

Get in touch