Expose OS resident set size (RSS) in allocator trace samples to find leaks

Many development and SRE teams struggle to find memory leaks because allocator accounting (what a language runtime reports) often diverges from what the OS actually keeps resident in memory (RSS), a gap that is growing as teams build polyglot apps with native modules and WebAssembly. This is especially painful in CI and cloud environments where memory regressions directly raise compute bills; with an addressable observability market of $24.0B (1.2M teams × $20K average annual spend) the operational and developer cost of hunting down leaks is non-trivial. You could build a tracing-enabled allocator sampler that annotates allocation call stacks with concurrent OS RSS samples, exposing per-sample RSS deltas and correlating them with traces and flamegraphs so developers can pinpoint allocations that increase real memory footprint. The product would include low-overhead, cross-platform sampling agents, integration into existing traces/metrics pipelines, CI gating and regression alerts, and UI features for historical comparison and root-cause attribution. This moment is attractive because observability consolidation is driving demand for unified traces, metrics and profilers, cloud and CI costs are rising, and the increase in native/JS hybrid stacks makes the RSS-versus-allocator gap more common; the project maps to a market score of 80/100 and revenue potential of 75/100 as a complementary signal to existing tools. The core strength is a unique, actionable signal—RSS at allocation-sample granularity—that addresses hard-to-diagnose leaks, but challenges include keeping sampling overhead very low, handling platform-specific RSS semantics, and achieving convincing ROI proofs for buyers in a medium-competition landscape.