DevOpsDays Singapore 2025: Top 8 Cloud Takeaways for Asia’s DevOps Leaders

The theme this year was the DevOps Meets AI: Transforming Engineering with Generative AI Tools. While the potential of GenAI is exciting, as DevOps practitioners we need to acknowledge several concerns that organizations must address. The entire AI technology stack is changing on a month-to-month basis, making it challenging for enterprises accustomed to more stable, long-term technology investments. 

1. Cloud-Native AI Ops is Here

In the opening keynote, Matthew Skelton emphasized the shift to AI-enhanced cloud operations, spotlighting platforms like Salesforce Hyperforce as enablers of scalable AI models across public cloud environments. DevOps teams are now embedding intelligent copilots into CI/CD pipelines, improving release cycles and boosting reliability in multi-cloud deployments.

2. CDKTF + AI: Redefining Infrastructure as Code

Vincent De Smet & Charles Martinot showed how using CDK for Terraform (CDKTF) simplifies managing hybrid cloud infrastructure — including Salesforce integrations. The takeaway: developers can now write secure, declarative Salesforce infrastructure code, enabling faster provisioning and better compliance across cloud ecosystems.

3. Observability in Action with ClickHouse Cloud

Jason Wong’s session demonstrated how ClickHouse Cloud supports scalable, high-speed logging for systems like Salesforce Service Cloud and Marketing Cloud. Paired with AI-driven anomaly detection, this enables:

• Real-time customer journey analytics
• Proactive performance monitoring across cloud platforms
• Faster incident resolution in Salesforce-powered environments

4. Optimizing Kubernetes on Cloud

Daniele Polencic presented AI-based Kubernetes resource scheduling to reduce cloud waste — highly applicable to Salesforce AppExchange partners deploying microservices. This aligns with Salesforce’s move toward containerization (Heroku → Hyperforce), where autoscaling and efficient memory allocation are vital to SLAs.

5. Cost Governance for Cloud DevOps

Sumit Srivastava discussed how InVideo uses Kubecost to manage GPU-heavy workloads on AWS. For organizations running Salesforce AI models or Einstein GPT extensions, this session emphasized:

• Cost attribution per dev team
• Budget caps tied to cloud workspaces
• Forecasting cloud resource needs for Salesforce customizations

6. Air‑Gapped Deployments for Sensitive Data

Ion Mudreac tackled a niche but critical area: on-premise, air-gapped AI workloads. For sectors using Salesforce Government Cloud or regulated customer datasets, hybrid CI/CD flows were key:

• Secure rollout of Einstein models behind firewalls
• Syncing air-gapped deployments with Salesforce API updates
• Caching dependencies and validating schema changes offline

7. DevSecOps for Salesforce Admins & Developers

In their talk, Lawrence Crowther & Craig Dent mapped DevSecOps evolution with clear relevance to Salesforce teams:

• Static scanning of Salesforce metadata and Apex code
• GitOps + CI/CD for Salesforce DX
• Guardrails built into Salesforce DevOps Center, integrating security policies early in the development cycle

8. AI-Powered Cloud Pipelines: Salesforce in Focus

Workshops like “Production-Grade LLM Apps” highlighted Salesforce’s AI direction:

• Using Einstein Copilot and Data Cloud in AI pipelines
• Deploying LLMs via MuleSoft APIs
• Monitoring AI-driven experiences with Salesforce-native observability tools and third-party cloud dashboards

Why This Matters for Salesforce Ecosystem in Asia

• Salesforce Hyperforce Expansion in Asia: Singapore and Japan now serve as regional data hubs
• Einstein GPT & Data Cloud: Requires DevOps teams to build scalable, AI-ready CI/CD frameworks
• Salesforce DevOps Center: Now widely adopted in Asia for CI/CD automation, with native Git integration