Articles and essays on AI evaluation, AI security, reliability, governance, application security, data quality and performance engineering.https://iamsatya.com/https://iamsatya.com/insights/agentic-ai-evaluation-framework/https://iamsatya.com/insights/agentic-ai-evaluation-framework/A practical framework for evaluating agents per layer, choosing metrics and thresholds, calibrating LLM judges against humans, and catching regressions across prompt and model changes.Thu, 01 Oct 2026 00:00:00 GMThttps://iamsatya.com/insights/query-engine-performance-regression/https://iamsatya.com/insights/query-engine-performance-regression/Baselines, noise, run counts, statistical comparison and per-query thresholds for catching Trino performance regressions in CI.Thu, 01 Oct 2026 00:00:00 GMThttps://iamsatya.com/insights/tpch-trino-query-plans/https://iamsatya.com/insights/tpch-trino-query-plans/How to read EXPLAIN and EXPLAIN ANALYZE in Trino, and what fragments, exchanges, join distribution, dynamic filters and statistics tell you.Wed, 30 Sep 2026 00:00:00 GMThttps://iamsatya.com/insights/indirect-prompt-injection-testing/https://iamsatya.com/insights/indirect-prompt-injection-testing/Where untrusted content enters an LLM pipeline, how to design planted-instruction tests with canaries, how to measure attack success, and how to verify defences actually hold.Tue, 29 Sep 2026 00:00:00 GMThttps://iamsatya.com/insights/validating-cdc-pipelines/https://iamsatya.com/insights/validating-cdc-pipelines/How to test incremental and change-data-capture pipelines for ordering, late and duplicate events, deletes, idempotent MERGE, watermarks and reconciliation windows.Sun, 27 Sep 2026 00:00:00 GMThttps://iamsatya.com/insights/agent-failure-handling-patterns/https://iamsatya.com/insights/agent-failure-handling-patterns/Retries, timeouts, fallbacks and recovery for AI agents, and the cases where a retry makes things worse.Sat, 26 Sep 2026 00:00:00 GMThttps://iamsatya.com/insights/tool-calling-reliability-testing/https://iamsatya.com/insights/tool-calling-reliability-testing/How to test whether an LLM picks the right tool, passes the right arguments, calls tools in a safe order and recovers from failure, with a pytest-style harness you can adapt.Thu, 24 Sep 2026 00:00:00 GMThttps://iamsatya.com/insights/security-findings-normalisation/https://iamsatya.com/insights/security-findings-normalisation/How to turn SAST, SCA, DAST and container scan output into one schema, with stable fingerprints, CWE and CVE mapping, normalised severity and expiring suppressions.Tue, 22 Sep 2026 00:00:00 GMThttps://iamsatya.com/insights/evals-are-the-new-tests/https://iamsatya.com/insights/evals-are-the-new-tests/Why a demo is not evidence, and what it takes to treat AI evaluation as a release gate rather than a research exercise.Sun, 20 Sep 2026 00:00:00 GMThttps://iamsatya.com/insights/agent-permissions-trust-boundaries/https://iamsatya.com/insights/agent-permissions-trust-boundaries/How to test what an AI agent is allowed to do, on whose behalf, and with whose data. Least privilege, confused deputies, MCP trust and approval gates, with a test matrix.Fri, 18 Sep 2026 00:00:00 GMThttps://iamsatya.com/insights/data-quality-rules-and-monitoring/https://iamsatya.com/insights/data-quality-rules-and-monitoring/Rule catalogues, severity and ownership, where checks run, simple anomaly baselines for volume, freshness and distribution, and alerting that people still read.Wed, 16 Sep 2026 00:00:00 GMThttps://iamsatya.com/insights/measuring-ai-reliability/https://iamsatya.com/insights/measuring-ai-reliability/Why a single pass proves little for non-deterministic systems, and how to measure consistency, set SLOs and error budgets, and choose thresholds you can defend.Tue, 15 Sep 2026 00:00:00 GMThttps://iamsatya.com/insights/verified-remediation-loop/https://iamsatya.com/insights/verified-remediation-loop/Closing the loop on vulnerabilities with ownership routing, severity-based SLAs, targeted re-scans, regression tests and metrics that leadership and auditors can trust.Sat, 12 Sep 2026 00:00:00 GMThttps://iamsatya.com/insights/governance-is-how-you-go-faster/https://iamsatya.com/insights/governance-is-how-you-go-faster/Risk tiers, model inventories and evidence trails are not bureaucracy. Done well, they are what lets an organisation say yes to AI.Fri, 28 Aug 2026 00:00:00 GMT