Healthcare / Cloud optimisation
VITech helped a large US healthcare platform save over $2M
How a single engineering team redesigned cloud infrastructure in under 3 months, cutting AWS costs by 60% and tripling report speed. All with zero disruption to HIPAA compliance.
$2M+
Saved in annual cloud costs
60%
Reduction in AWS spend
3×
Faster nightly reporting
At a glance
About the client
Healthcare / Cloud optimisation
Key technologies
Java
.NET
Spring
AWS ECS
Lambda
RDS
S3
SQS
Challenge
Soaring AWS costs, 18-hour report cycles, and outdated architecture blocking scalability.
Results
Reduced AWS costs from $3.6M to $1.5M annually, cut report generation from 18 to 6 hours, scaled without new infrastructure, and maintained full HIPAA compliance.
High costs, slow insights
The challenge
Over 2-3 months, our team cut AWS costs by 60% and tripled report speed while maintaining HIPAA compliance.
BEFORE
Soaring AWS costs at $3.6M/year with no clear optimization path
AFTER VITECH
AWS costs cut from $3.6M to $1.5M annually — $2M+ savings
BEFORE
18-hour nightly report generation cycles
AFTER VITECH
Reports ready in 6 hours — 3× faster
BEFORE
All workloads ran on high-resource compute (4 CPUs, 8 GB) regardless of file size
AFTER VITECH
Dual-path routing: 99% of files on lightweight compute, only 1% use high-resource lanes
BEFORE
2PB+ of rarely accessed S3 data continuously consuming resources
AFTER VITECH
Stale S3 files generated on demand — data refreshed only when needed
BEFORE
Architecture blocking new client onboarding and compliance rollouts
AFTER VITECH
Scalable for new clients and regulatory reports — no extra infrastructure required
The solution
We’ve done three key initiatives to optimize the platform
and tackle the client’s challenges.
Optimization of compute
& smart scheduling
Past years no longer trigger full nightly recalculations
Full-year views now update only when new patient data comes in
Archived data are served from cache, so the system stops reprocessing it
Optimization of storage strategy
Deleting infrequently used data from Amazon S3 storage.
Making stale files generate on demand, so data is refreshed only when needed.
Optimization of resource allocation
99% of small files now run on lightweight compute (2 CPUs, 4 GB RAM).
Large files (1%) get sent down to high-resource lanes (4 CPUs, 8 GB RAM).
Achieved 3× faster nightly processing and reduced unnecessary compute usage
