Tune and deploy Gemini with Vertex AI and ground with Cloud databases

Introduction to the session with Bazir Fat and Balanarsimha discussing the use of Vertex AI for Google's GeminiRo language model.

Discussion about concerns in online newspapers such as declining customer satisfaction and increased churn rate.

Exploring the use of generative AI for improving website navigation and content summaries in online newspapers.

Detailed discussion on building generative AI applications, evaluating LLM models, crafting prompt templates, and testing performance.

Demonstration on using Vex AI for crafting prompt templates, evaluating different LLM models, and deploying a subhead summaries generator for news articles.

Explanation on tuning models using Vertex AI's fully managed service, Vex AItuning, for optimizing model performance.

Discussion on evaluating tuned models using different metrics and comparing the performance before and after tuning.

Overview on monitoring and evaluating generative AI models in production using Vertex AI computation-based evaluation and auto side-by-side evaluation.

Process of deploying generative AI applications using a jumpstart solution with technology components like Vertex AI and Cloud SQL for Postgres.

Exploration of observability features in production, including SSL encryption, IM-based authentication, private IP setup, building, deploying applications in GKE, and observability tools like query insights and data cache metrics.