Overview & philosophy
Michael believes the data path and the message path should be as visible in your repo as the HTTP layer. When those layers are opaque, teams ship features fast—and then spend quarters untangling production behavior nobody modeled.
He has led remediation on systems where “it worked in QA” failed because volume, timing, and locking were never simulated. His default response is not more logging for its own sake, but better invariants: idempotency keys, deduplication windows, explicit outboxes, and consumer concurrency limits that match real broker settings.
He writes for the engineer who gets paged at 2 a.m.: what broke, how to confirm it, how to mitigate, how to prevent recurrence—captured in tickets and runbooks, not tribal knowledge.
“Michael is who we bring in when the dashboard is green but customers are still angry—he finds the story in SQL waits and queue depth.” — Platform director, high-volume e-commerce (anonymized)
Technical depth
A concise map of where Michael spends deep time. He will happily pair with your DBA or SRE; he does not throw problems “over the wall.”
SQL Server & Azure SQL
Async & resilience
.NET service architecture
- Worker services, outbox pattern, and saga-style coordination where business rules require it.
- Containerized workloads when process isolation or density demands it.
- BenchmarkDotNet for proving claims; load tests scoped to realistic mixes.
Tools & platforms (representative)
Representative project types
Illustrative categories of backend work Michael has repeatedly owned—use them to compare against your open problems.
Throughput & latency programs
Profiling hot endpoints, batching vs. streaming trade-offs, and cache strategies that do not silently corrupt business rules.
Queue migration & dual-write
Moving between brokers or patterns with reconciliation jobs, idempotent consumers, and cutover playbooks.
Incident hardening
Postmortems that produce code and config changes—not slide decks. Alert noise reduction and SLO-aligned dashboards.
Engagement models
Michael is often engaged for focused intensity—performance remediation, pre-launch hardening, or architecture for a new high-scale surface—then optionally stays for steady-state improvements.
| Model | Minimum | Best when… |
|---|---|---|
| Dedicated monthly | Often 3+ months | Ongoing platform work with a mix of features and reliability investments. |
| Performance / reliability sprint | 2–8 weeks | You have a measurable target: P95 latency, error rate, or cost per transaction. |
| Architecture advisory | Short intensive | You want a second opinion on topology, data model, or queue design before scaling the team. |
Metrics, traces, and representative workloads captured before large code changes.
Feature flags, canary releases, and rollback scripts where the blast radius warrants them.
Before/after numbers in the ticket, runbook updates, and alert thresholds tied to reality.
Ways of working with your team
Engineering managers: Michael translates technical risk into delivery trade-offs: what slips if we skip hardening, what accelerates if we narrow scope for two sprints.
DBAs / data teams: He brings reproduction scripts, proposed index changes with rollback, and clear questions about retention and RPO/RTO—not vague “DB slow” tickets.
Frontend teams: He cares about contract stability and pagination semantics so UI engineers are not fighting accidental N+1 or ambiguous error payloads.
Share your slowest endpoint (URL + rough RPS), top 3 tables by growth, and current broker settings—Michael can usually propose a sensible first-week plan from that alone.
Fit, boundaries & quick answers
- • Need throughput, back-pressure, and data integrity together.
- • Have SQL pain alongside .NET services—not one or the other in isolation.
- • Want benchmarks and plans, not vibes.
- • Primary need is marketing-site front-end only.
- • Stack is exclusively non-Microsoft data with no .NET integration surface.
- • Expect greenfield React ownership without meaningful API/data work.
Will Michael rewrite our entire database?
Only when justified by evidence and phased with rollback. Prefer incremental improvements with measured impact.
Can he join production incidents?
Yes, when engaged with clear responsibilities and runbook access—often paired with your on-call lead for knowledge transfer.
Does he document what he changes?
Yes—runbooks, architecture notes, and ticket comments aimed at the next on-call engineer.
Need Michael on your critical path?
Send traffic patterns, pain metrics, and SLAs—we’ll reply with a focused plan and start window.