Thank you for sending your enquiry! One of our team members will contact you shortly.
Thank you for sending your booking! One of our team members will contact you shortly.
Duration 35 hours
Course Outline
1. LLM Architecture and Core Techniques
- Evaluating Decoder-Only (GPT-style) versus Encoder-Decoder (BERT-style) model architectures.
- Analyzing Multi-Head Self-Attention, positional encoding, and dynamic tokenization in detail.
- Exploring advanced sampling methods, including temperature, top-p, beam search, logit bias, and sequential penalties.
- Conducting a comparative analysis of leading models such as GPT-4o, Claude 3 Opus, Gemini 1.5 Flash, Mistral 8×22B, LLaMA 3 70B, and quantized edge variants.
2. Enterprise Prompt Engineering
- Implementing prompt layering strategies involving system, context, user prompts, and post-processing.
- Applying Chain-of-Thought, ReACT, and auto-CoT techniques with dynamic variables.
- Designing structured prompts using JSON schemas, Markdown templates, and YAML function-calling.
- Implementing mitigation strategies for prompt injection, such as sanitization, length constraints, and fallback defaults.
3. AI Tooling for Developers
- Reviewing and comparing the utility of GitHub Copilot, Gemini Code Assist, Claude SDKs, Cursor, and Cody.
- Applying best practices for integrating AI tools into IntelliJ (Scala) and VSCode (JS/Python) environments.
- Performing cross-language benchmarks for coding, test generation, and refactoring tasks.
- Customizing prompts per tool through aliases, contextual windows, and snippet reuse.
4. API Integration and Orchestration
- Executing end-to-end implementations of OpenAI Function Calling, Gemini API Schemas, and Claude SDKs.
- Managing rate limits, error handling, retry logic, and billing metering effectively.
- Constructing language-specific wrappers:
- Scala: Utilizing Akka HTTP
- Python: Leveraging FastAPI
- Node.js/TypeScript: Using Express
- Incorporating LangChain components, including Memory, Chains, Agents, Tools, multi-turn conversation logic, and fallback chaining.
5. Retrieval-Augmented Generation (RAG)
- Parsing technical documentation formats (Markdown, PDF, Swagger, CSV) using LangChain/LlamaIndex.
- Applying semantic segmentation and intelligent deduplication techniques.
- Working with various embedding models, including MiniLM, Instructor XL, OpenAI embeddings, and local Mistral embeddings.
- Managing vector stores like Weaviate, Qdrant, ChromaDB, and Pinecone, focusing on ranking and nearest-neighbor tuning.
- Implementing low-confidence fallback mechanisms to alternative LLMs or retrievers.
6. Security, Privacy, and Deployment
- Implementing PII masking, prompt contamination control, context sanitization, and token encryption.
- Establishing prompt and output tracing through audit trails and unique IDs for every LLM call.
- Configuring self-hosted LLM servers (Ollama + Mistral) with GPU optimization and 4-bit/8-bit quantization.
- Deploying on Kubernetes using Helm charts, autoscaling strategies, and warm start optimizations.
Hands-On Labs
- Prompt-Based JavaScript Refactoring
- Executing multi-step prompting workflows: detect code smells, propose refactors, generate unit tests, and create inline documentation.
- Scala Test Generation
- Generating property-based tests using Copilot versus Claude, measuring coverage and edge-case efficacy.
- AI Microservice Wrapper
- Building a REST endpoint that processes prompts, forwards them to LLMs via function-calling, logs outcomes, and handles fallback logic.
- Full RAG Pipeline
- Constructing a complete workflow: simulated documents, indexing, embedding, retrieval, and search interfaces with ranking metrics.
- Multi-Model Deployment
- Setting up a containerized environment with Claude as the primary model and Ollama as a quantized fallback, monitored via Grafana with defined alert thresholds.
Deliverables
- A shared Git repository housing code samples, wrappers, and prompt tests.
- A benchmark report detailing latency, token costs, and coverage metrics.
- A pre-configured Grafana dashboard for monitoring LLM interactions.
- Comprehensive technical PDF documentation and a versioned prompt library.
Troubleshooting
Summary and Next Steps
Requirements
- Proficiency in at least one programming language, such as Scala, Python, or JavaScript.
- Working knowledge of Git, REST API design, and CI/CD workflows.
- Familiarity with core Docker and Kubernetes concepts.
- A strong interest in leveraging AI/LLM technologies within enterprise software engineering contexts.
Target Audience
- Software Engineers and AI Developers
- Technical Architects and Solution Designers
- DevOps Engineers focused on AI pipeline implementation
- R&D teams exploring AI-assisted development methods
Testimonials (2)
The course was very useful, and the trainer was clear, well-prepared, and engaging. I liked the fact that it was very practical with labs and real use cases. Overall, it was a valuable training experience.
Mattia Dettori - MFM INVESTMENT Ltd Italian branch
Course - LLM Engineering Bootcamp
1. The section on the constraints, tests, and guardrails was cool. 2. Francesco knows the topic very well, and he’s a nice teacher. 3. Overall, I really enjoyed the course, and I learnt a lot.
