Скачать или смотреть Optimizing LLM Workflows with DSPy and Gemma: Insights from DevFest Hong Kong 2025

In this exclusive interview from the DevFest Hong Kong 2025 Creator Studio, Matthias Baertens, a Cloud Architect based in Singapore, dives deep into his latest work on Google Cloud.

Matthias discusses building an advanced LLM-based system designed to streamline customer support routing and technical detail extraction. Learn how he leveraged DSPy for structured prompt optimization and experimented with fine-tuning Google's Gemma open models to achieve a 90% performance rate.

Key Topics Covered:
[00:33] Automating ticket classification and metadata extraction for Cloud Architects.
[01:46] Using the DSPy library for structured prompt optimization.
[01:53] Fine-tuning Gemma open models for specialized tasks.
[03:07] Reflections on the Hong Kong tech talent ecosystem and startup opportunities.
[04:31] Advice for creators: Focusing on the Agentic Space (ADK and Vertex AI).

Other DevFest Video    • DevFest 2025 Hong Kong  

More intresting interview in Creator Studio    • Creator Studio - DevFest 2025 Hong Kong  

#devfest2025 #MatthiasBaertens #googlecloud #DSPy #GemmaLLM #vertexai #promptengineering #AIAgents #machinelearning #gdgcloudhk #ADK #cloudarchitect .