Charlie Smith

For years, SQL performance reviews depended on experience and vigilance. Senior DBAs manually catching missing indexes, inefficient joins, and non-scalable patterns before they reached production. It worked, but it did not scale. It relied on tribal knowledge and created review bottlenecks. I decided to turn that expertise into a system. Using Cursor and the Model Context Protocol, I built a context-aware AI agent dedicated to SQL performance validation. The difference is not in syntax checking. It is in context. The agent has access to our actual schema, foreign keys, indexes, and data relationships. It evaluates queries the way a database engineer would, identifying when a join will not scale, when a covering index is missing, and when execution plans will degrade under real workloads. Today, every SQL change runs through this layer before human review. Feedback is immediate, objective, and grounded in our data model. Review cycles are shorter. Discussions are more architectural and less corrective. Performance becomes a shared responsibility rather than a last-minute gate. The next step is embedding this capability directly into developers’ IDEs. Not as a policing tool, but as a real-time guardrail. The goal is to shift database expertise left and make scalable query design the default behavior. Generative AI that writes code is impressive. Context-aware AI that operationalizes domain expertise is transformative. The real opportunity is not replacing engineers. It is encoding hard-earned knowledge into intelligent systems that elevate the entire engineering organization. #AI #GenerativeAI #Database #SQL #DataEngineering #DevTools #ShiftLeft #EngineeringLeadership #MCP #SoftwareEngineering #pango

7

4 Comments

🚀 Ever spent hours analyzing slow queries and figuring out the right indexes? I just released a new video showing how to automate database performance tuning with IQToolkit Analyzer! In this demo, I take a PostgreSQL query with a slow Sequential Scan (20 seconds!) and use AI to generate the perfect index strategy - cutting execution time dramatically. 💡 What you'll learn: ✅ How to identify performance bottlenecks with EXPLAIN ANALYZE ✅ Using AI to recommend optimal index strategies ✅ Real-world example with measurable results Whether you're a DBA, backend engineer, or just curious about AI-powered database optimization, this 17-minute walkthrough shows practical automation in action. Check it out and let me know what you think! 👇 https://lnkd.in/es_9fWtP #PostgreSQL #DatabaseOptimization #AI #DatabasePerformance #DBA #IQToolkit #OpenSource

6

Documenting for future me. If you are trying to CREATE VECTOR INDEX in the pre-release of SQL Server 2025, documentation was to turn on Trace Flags 466, 474, and 13981. This will cause an error message of 343 on SQL Server 2025 RTM. So, be sure to turn these off. DBCC TRACEOFF(466, 474, 13981, -1)

8

Snowflake credits aren’t just compute. They hide in storage too. Unused dev/test databases, stale tables, and long Time Travel or Fail-Safe settings silently drain budgets every month. Many teams never notice until credit burn rates spike. A recent Snowflake case study showed that 40% of storage in a large enterprise came from inactive data objects. That’s not innovation — that’s waste. At Anavsan, we built an AI-driven way to stop it. Our platform automatically scans your Snowflake environment, identifies idle tables, clones, and retention bloat — and helps you reclaim credits within weeks, not months. -- Pinpoint inactive storage -- Automate cleanup and governance using a secure workspace -- Reuse saved credits for new workloads Read the full blog: https://lnkd.in/eXUFNgkd Because in FinOps, every reclaimed credit fuels what really matters — growth, not waste. #FinOps #Snowflake #CloudOptimization #DataEngineering #Anavsan #CloudCostOptimization

2