Free Cloud Egress Audit · No call required
Find the hidden transfer tax in your cloud bill.
Your intake goes to me — and I read your whole stack.
I work where AI meets HPC, with a view that runs from transistor-level architecture to cloud-scale inference. That full-stack span is how I find cost leaks others miss — in the silicon, the serving layer, and the system design at once.
Transistor→Kernel→GPU→Cluster→Cloud
Vikas Chamarthi · Founder, NavyaAI
LinkedInTrack record
M.S. Electrical & Computer Engineering, UNC Charlotte
10–15%
of cloud bill is egress
7.5–27%
of total bill recoverable
$6.3K→$0
one cross-region path
Modeled at list price.
Egress guide
Selected work — benchmarks & reports
We publish our numbers so you can check our work, on systems depth and on cost saved.
The hidden cloud tax
Field guide
Cloud egress cost optimization: where the bytes really cost
Internet-out, NAT, cross-AZ, cross-region, CDN cache-fill, and Supabase realtime fanout — and the architecture moves that cut each one.
Read the write-up
Token prices −99.7%, bills up 3×
Report
Why unit prices collapsed but your cloud bill tripled
Per-unit cost keeps falling while bills climb. The gap is volume, architecture, and transfer — not the rate card.
Read the report
AI Economics 2026
Report
Token collapse and the sustainability question
Our 2026 read on infrastructure economics: deflation, margin pressure, and which deployment models actually survive.
Read the report
Engineering
70% faster, 75% less memory
Threads Beat Multiprocessing for RAG: 70% Faster, 75% Less Memory
Read the write-up· 12 min read
Engineering
Rust vs Python, measured
Embedding Rerank Gateway: Rust vs Python Cost and Performance
Read the write-up· 10 min read
Engineering
Same model, lower bill
Python vs Rust for Transformers: Performance and Cost
Read the write-up· 25 min read
Sound familiar?
The bill grew. The product didn't.
- NAT and private-path markup
Private-subnet workloads pay for internet egress, then again for NAT processing, then again for cross-AZ if the path isn't local. NAT alone can add 44–61% on a heavy outbound workload.
- Cross-AZ and cross-region chatter
Chatty services and multi-region replication quietly become multi-thousand-dollar line items. On AWS, internet-out plus regional transfer is ~90% of transfer spend.
- Realtime and serverless fanout
Supabase realtime, webhooks, and Trigger jobs turn one row change into many deliveries. A single feature can run $3,000+/month in fanout you'd never see coming.
Illustrative breakdown at list price. Most transfer cost is architectural, not the per-GB rate — NAT, cross-AZ, and realtime fanout are where it hides. Read the egress guide.
Audit scope
Eleven places transfer cost hides
The intake qualifies your stack, spend, and biggest transfer driver. Then I inspect the paths that actually move the bill.
- Public internet egress by service
- Cross-AZ traffic share
- Cross-region replication and reads
- NAT gateway processing
- Private endpoint / PrivateLink processing
- CDN cache-fill and lookups
- Supabase realtime fanout
- Edge Function / serverless invocations
- Websocket payload size
- Backups, restores, and exports
- Polling vs. webhook / outbox patterns
How it works
Intake to findings in three steps
- 01
Share your stack shape
Primary cloud or stack, monthly spend, and your biggest transfer driver. No billing access required to start.
~3 minutes
- 02
I map the avoidable paths
I match your shape against known leak patterns — NAT, cross-AZ, cross-region, CDN cache-fill, realtime fanout — and rank what to inspect first.
Written reply
- 03
Get the right next step
A ranked list of avoidable transfer paths, a rough savings range, and recommended fixes by complexity and payback.
Your call from there
Go deeper on egress
The field guide, the Supabase-specific playbook, and the AI inference audit if that's your bigger line item.
FAQ
Common questions before requesting the audit
Answers written and last reviewed by Vikas Chamarthi, Founder of NavyaAI, on .
What is a cloud egress audit?
It's a review of where data-transfer cost actually accumulates in your stack — not just internet egress, but NAT gateway processing, cross-AZ and cross-region traffic, private-endpoint data processing, CDN cache-fill, and realtime/serverless fanout. I find the top avoidable paths before you renegotiate rates or buy more capacity.
Who is the free egress audit for?
Teams on AWS, GCP, or Azure where transfer is a visible line item, and teams on modern stacks (Supabase, Neon, Vercel, Trigger.dev) where realtime or serverless fanout is quietly scaling cost. Usually most useful above ~$10K/month of cloud spend, where the leaks are large enough to matter.
Why is egress more than just internet bandwidth?
On a typical bill, transfer is spread across many SKUs: internet-out, cross-AZ (~$0.01/GB each direction), cross-region (~$0.02/GB), NAT processing (~$0.045/GB plus hourly), private-endpoint processing, and CDN cache-fill. NAT alone can add 44–61% on top of a heavy outbound workload. Most of the cost is architectural, not the per-GB rate.
How does Supabase or Trigger.dev burn egress at scale?
Realtime fanout multiplies one row change into many subscriber deliveries. A single feature with 500 subscribers and 2KB payloads can run ~$3,000+/month in realtime message overage plus egress — entirely from event topology, not traffic you'd notice. The fixes are channel scoping, payload minimization, batching, and outbox patterns.
Is this a disguised sales call?
No. I reply with written findings — your top avoidable transfer paths and a rough savings range — whether or not we ever talk. A call only happens if it's genuinely the right next step and you want one.
Free · 3-minute intake
Find your top avoidable transfer paths.
Written findings, no call required. The worst case is you confirm your transfer is already tight.
No credit card · No call required · Work email only