Master the integration of LLMs into your IS : Training for IS Developers and Architects
Modalities
Master the integration of LLMs into your IS. This programme, which covers the theoretical and practical aspects of the field of LLM integration, lasts a total of 20 hours and is delivered in the form of interactive online modules or face-to-face Intra-Company sessions. It combines theoretical sessions, practical demonstrations and collaborative workshops developing real-life use cases, to ensure a thorough understanding and practical application of the concepts covered.
Duration: 21 hours (3 days)
Public
Developers and architects wishing to acquire the skills needed to integrate Large Language Models (LLMs) into their future developments.
Trainer
Boris Yepmo
- AI, Big Data and Machine Learning expert, 3 IL Limoges IT Design and IT Management Engineer.
- Master’s degree in Computer Systems and Networks
- BIG Data Specialisation – University of California San Diego
- HEC Mentor Teacher
- Professor of Artificial Intelligence ESCP
- Lecturer at the University of Lorraine
- Head of the Global IT Management Major at EPITA
- Teacher Information System Management ESSCA
Learning objectives
- Understanding the fundamentals of LLMs : Acquire an in-depth knowledge of the key concepts associated with large-scale language models, including their architecture, operation and practical applications.
- Master the associated tools and frameworks : Become familiar with the main tools and frameworks used for the development and integration of LLMs, such as LangChain and the generative AI APIs available on platforms such as Google Cloud.
- Applying prompt engineering techniques : learning how to design effective prompts to optimise the performance of LLMs in various application contexts.
- Integrating LLMs into existing IS architectures : Develop the ability to effectively integrate LLMs into existing information systems, taking into account technical constraints and specific business needs.
- Ensuring ethical and secure use of LLMs : Understanding the ethical issues associated with the use of LLMs and implementing measures to guarantee data security and compliance with current regulations.
Acquired skills in mastering the integration of LLMs into your IS
- Analysis and selection of suitable models : Ability to evaluate and select the most appropriate LLMs according to specific project needs and technical constraints.
- Development and deployment of solutions based on LLMs : Competence in the design, development and deployment of applications incorporating LLMs, using frameworks such as LangChain.
- LLM performance optimisation : Ability to apply prompt engineering techniques to improve the relevance and efficiency of responses generated by LLMs.
- Integration into complex IS environments : Mastery of the integration of LLMs into existing information system architectures, ensuring optimum interoperability and efficient resource management.
- Management of ethical and security aspects : Knowledge of best practices to ensure responsible use of LLMs, in compliance with ethical standards and data security requirements.
Pedagogical objectives
1. Introduction to LLMs and Generative AI (3 hours)
Fundamental concepts: Understanding the principles of artificial intelligence, machine learning and large language models.
History and evolution: Study the evolution of LLMs, from the first models to recent advances.
Practical applications: Explore the use cases of LLMs in various sectors.
2. LLM architecture (3 hours)
Internal mechanisms: Taking a closer look at Transformer-type architectures, attention mechanisms and tokenisation.
Model training: Understanding pre-training, fine-tuning and supervised and unsupervised learning techniques.
3. Tools and Frameworks (3 hours)
LangChain: Discover this open source framework for developing applications powered by language models.
API and cloud services: Using services such as Google Cloud’s PaLM API to integrate LLMs into applications.
4. Prompt Engineering (2 hours)
Designing effective prompts: Learn how to formulate prompts to obtain precise responses from LLMs.
Advanced techniques: Explore Chain-of-Thought and Retrieval Augmented Generation (RAG) approaches to improve model performance.
5. Integrating LLMs into Applications (3 hours)
Developing chatbots: Creating multi-turn chat applications using LLMs and frameworks such as LangChain.
Customizing models: Adapting LLMs to the specific needs of the company using fine-tuning.
6. Ethical and Security Considerations (2 hours)
Bias and fairness: Identifying and mitigating bias in language models.
Data security: Ensuring confidentiality and data protection when using LLMs.
7. Practical Workshops and Projects (4 hours)
Case studies: Working on practical cases to reinforce the skills acquired.
Collaborative projects: Developing applications incorporating LLMs in teams to encourage learning by doing.