Ways to work together
Four engagement types
Each describes when it fits, what it covers, and what you get at the end. None of these are fixed-price packages — scope is agreed before work starts.
INVESTIGATION
JVM performance audit
When it fits: you have a specific performance concern — high p99, a throughput ceiling, unexplained GC pauses, rising CPU — but no clear diagnosis yet.
What it covers: workload analysis, profiling, allocation and GC review, contention analysis and JIT/compilation behaviour, against a representative workload.
What you get: a written diagnosis of the actual bottleneck(s), evidence behind each finding, and a prioritised list of remediation options — not a guaranteed percentage improvement.
ARCHITECTURE
Low-latency design review
When it fits: you're designing or reworking a latency-sensitive component and want a second, specialised opinion before or during implementation.
What it covers: execution model, state ownership, communication paths, persistence boundaries and failure semantics, reviewed against your actual latency and throughput targets.
What you get: a written review — risks, trade-offs and concrete recommendations tied to your constraints.
PRODUCTION
Performance intervention
When it fits: something in production has regressed or hit a wall — a latency regression, throughput collapse, memory pressure, high CPU, lock contention or a scaling limit — and you need focused help resolving it.
What it covers: targeted diagnosis on the affected path, working from your existing telemetry and profiling where available, adding instrumentation where it's missing.
What you get: root-cause findings and a concrete fix or remediation plan your team can implement and validate.
ENABLEMENT
Advanced engineering training
When it fits: your team is capable but hasn't built deep intuition for JVM internals, concurrency or performance-conscious design, and would benefit from structured, hands-on sessions.
What it covers: JVM internals, concurrency and memory ordering, profiling and benchmarking methodology, and performance-conscious architecture — tailored to your codebase and team level.
What you get: working material and exercises your team keeps, built around problems close to your own systems.