reliquary.cloud
 Start a Project →
// CASE_STUDIES

Shipped. Measured. Still running.

Every project below has a public-safe summary. Detailed results are available under NDA on request.

// CLIENT

Kepler Health

HIPAA-compliant telehealth platform

Problem

Legacy monolith on EC2 with 14-minute deploy cycles, four-hour incidents, and compliance debt blocking their Series B.

Approach

Twelve-week engagement. Migrated to EKS with a Backstage portal, OpenTelemetry pipeline, and a HIPAA-aware landing zone in AWS. Trained their team through pairing and tabletop exercises.

Results
  • 14 min → 4 min deploy cycle
  • 4h → 12 min MTTR on P1
  • SOC 2 Type II audit passed first time
  • $38k/month infra savings
// CLIENT

Substrate Trading

Real-time risk and post-trade platform

Problem

Co-located infrastructure approaching end-of-life. Regulatory requirements tightening around observability and audit trail.

Approach

Hybrid migration: kept latency-sensitive matching on-prem, moved analytics and reporting to GCP with a private interconnect. Built the observability stack on OpenTelemetry + Tempo + Grafana.

Results
  • Zero post-migration downtime
  • 99.995% platform uptime maintained
  • MiFID II reporting latency within SLA
  • 41% reduction in ops headcount
// CLIENT

Lattice Labs

Developer productivity SaaS (Series B)

Problem

Customer base grew 10x in eighteen months. Infra cost grew 11x. Every engineering hire was spending 30% of week one on environment setup.

Approach

Built an internal developer platform on Kubernetes with Crossplane for cloud resources. Unified staging/prod pipeline with Argo CD. Wrote the developer onboarding runbook that now gets hires productive in days one.

Results
  • Onboarding: 2 weeks → 2 days
  • Infra cost/customer down 63%
  • Deploy frequency 4x/week → 27x/day
  • DORA elite status achieved
// CLIENT

Heliostat Media

Video ingestion and streaming

Problem

Traffic spikes during live events caused cascading autoscaling failures. Customer-facing outages during the moments that mattered most.

Approach

Rebuilt the ingestion pipeline on Kinesis + ECS Fargate. Introduced Karpenter for elastic compute. Built predictive autoscaling tied to scheduled event calendars.

Results
  • Zero outages through peak streaming quarter
  • 68% reduction in over-provisioning cost
  • 5-minute capacity pre-warm for scheduled events
  • Live event viewership up 2.3x