https://engineering-leadership.hinshelwood.com/engineering-notes/ Recent content in Engineering-Notes on Engineering Leadership in AI & Software Hugo en Wed, 04 Jun 2025 09:00:00 +0000 https://engineering-leadership.hinshelwood.com/engineering-notes/building-a-resilient-token-server-engineering-for-flow-fault-tolerance-and-speed/ Thu, 08 May 2025 09:00:00 +0000 https://engineering-leadership.hinshelwood.com/engineering-notes/building-a-resilient-token-server-engineering-for-flow-fault-tolerance-and-speed/ Aiming for a resilient, fast, and fault-tolerant token counting system, the author replaced fragile server restarts with a batch-wide server lifecycle, added retry logic for transient failures, and implemented a local fallback to ensure uninterrupted processing. These changes improved reliability, reduced downtime, and provided clear logs for troubleshooting. Development managers should focus on building systems that handle real-world failures gracefully, prioritize flow, and include observability and fallback mechanisms from the start. https://engineering-leadership.hinshelwood.com/engineering-notes/convert-legacy-projects-and-asp-net-mvc-apps-to-sdk-style-with-confidence/ Thu, 29 May 2025 09:00:00 +0000 https://engineering-leadership.hinshelwood.com/engineering-notes/convert-legacy-projects-and-asp-net-mvc-apps-to-sdk-style-with-confidence/ Upgrading legacy .NET projects to SDK-style makes your codebase easier to maintain, improves build and CI/CD integration, and prepares you for future .NET upgrades. Converting class libraries is straightforward and highly recommended, while web apps require more care and are best handled with the MSBuild.SDK.SystemWeb package if you want SDK-style; otherwise, you can leave them in the old format. Start by converting class libraries now to simplify your engineering system and reduce technical debt. https://engineering-leadership.hinshelwood.com/engineering-notes/how-i-used-generative-ai-to-transform-site-tagging-and-categories/ Thu, 15 May 2025 09:00:00 +0000 https://engineering-leadership.hinshelwood.com/engineering-notes/how-i-used-generative-ai-to-transform-site-tagging-and-categories/ Migrating a large, legacy blog to Hugo enabled the use of generative AI for automated tagging and categorisation, significantly improving content discoverability and editorial consistency while reducing manual effort. The system combines AI-driven suggestions with human oversight, using multi-factor scoring, penalty logic, and transparent reasoning to ensure quality and accountability. Development managers considering similar automation should maintain human control over final decisions and leverage AI to streamline, not replace, editorial processes. https://engineering-leadership.hinshelwood.com/engineering-notes/leveraging-ai-embeddings-for-related-content-classification/ Wed, 04 Jun 2025 09:00:00 +0000 https://engineering-leadership.hinshelwood.com/engineering-notes/leveraging-ai-embeddings-for-related-content-classification/ Switching from Hugo’s basic related content feature to an AI embeddings-based approach enabled much more accurate and meaningful content recommendations, improving both user navigation and AI discoverability. The solution was cost-effective, scalable, and reduced computational overhead by caching similarity scores and storing embeddings for reuse. Development managers can consider using AI embeddings for smarter content linking, which can enhance user experience and support future automation or classification needs.