Building Trusted AI Engineering
AI is changing how we build software. But how do we build systems we can trust?
Join Open Source Singapore Meetup to explore AI-native engineering, open-source frameworks for production AI, and real-world lessons from LLM performance. Discover how to build faster, more reliable software with AI while maintaining engineering discipline and control.
Agenda:
6:30 PM: Registration
6:50 PM: Welcome and opening remarks
7:00 PM: Building AI-Native Engineering Teams Without Losing Engineering Discipline
7:25 PM: LLM4S 1.0: From Idea to Production AI in Scala/JVM
7:50 PM: Fast Models, Slow Systems: Lessons from Benchmarking an LLM
8:15 PM: Networking
8:30 PM: Event Ends
1. Building AI-Native Engineering Teams Without Losing Engineering Discipline
Speaker: Ilya Verbitskiy
Ilya Verbitskiy is a software engineer and architect with over 20 years of experience building cloud platforms, security systems, and developer tools. His current work focuses on AI products, agent-based software, and practical approaches to reliable AI-native engineering.
AI-native engineering is not about replacing engineers with prompts. It is about redesigning the delivery system so humans and AI work together with clear intent, strong constraints, and rigorous verification. This talk shows how small, senior teams can use AI to ship products in weeks without accumulating architectural debt. We will explore why faster code generation makes engineering discipline more important, not less, and how teams can avoid the common failure mode of plausible but incoherent AI-generated code.
2. LLM4S 1.0: From Idea to Production AI in Scala/JVM
Speaker: Kannupriya Kalra
Kannupriya Kalra is the Co-Founder of the LLM4S Foundation and creator of LLM4S, an open-source generative AI framework for Scala and the JVM.
From an open-source idea to a production-ready AI framework, LLM4S has evolved rapidly over the past 18 months. Kannupriya will share the journey toward LLM4S 1.0, including lessons learned and why Scala's type system and functional programming foundations matter for reliable AI applications. Through live demos, explore LLM integration, tool calling, MCP, agent orchestration, observability, and practical approaches to building production AI applications in Scala and the JVM.
3. Fast Models, Slow Systems: Lessons from Benchmarking an LLM
Speaker: Matthew Alex
Matt is an engineer building high-performance systems and learning about ML inference. He enjoys conversations about economics, the future of AI, and movies.
How many users can a Mac mini running an LLM handle? Early tests looked promising, until requests started taking ten seconds and the dashboard still showed no queue. Drawing on his first experiments with LLM serving, Matt shares the testing mistakes, confusing results, and lessons learned about measuring performance when requests pile up.
Open Source Singapore:
Open Source Singapore (formerly Free and Open Source Meetup) is a community dedicated to promoting awareness, knowledge-sharing, and adoption of open source technologies. We bring together developers, enthusiasts, and industry professionals to exchange ideas, collaborate, and champion the values of openness, transparency, and innovation in Singapore’s tech ecosystem.
Venue Sponsor