00CraneDev
Production software, built to be maintained.
We take systems from architecture to production, then stay on them. No handover to a team that has never seen the code.
Cut p95 write latency by moving the ledger to append-only
a3f19c2 · cranedev/ledger-core · 2h ago
01Proof
Numbers from production, not from a deck.
42
Systems in production
180ms
Median p95 after migration
99.98%
Uptime across managed services
9 yrs
Median engineer experience
02Services
How we work, and what each engagement leaves you with.
- 2–3 weeks
Architecture and discovery
We map the system you have against the one you need, and write down why.
What you get
A written architecture, a risk list, and a sequenced plan you own.
- 8–16 weeks
Build to production
A senior team takes the plan to a running system, in your repo from day one.
What you get
A deployed service, tests, runbooks, and the CI that ships it.
- 6–14 weeks
Rescue and migration
For systems that work but cannot be changed safely any more.
What you get
Incremental migration with a rollback at every step, no big-bang cutover.
- Ongoing
Run it with you
We stay on the system after launch instead of handing it to strangers.
What you get
On-call rotation, monthly reliability review, and a shrinking backlog.
03Selected work
Systems we built, and the numbers that moved.
Every card opens the same three lines: the problem, what we built, and what it measured afterwards.
Fintech2025
Northwind Pay
Cut p95 write latency from 1.4s to 180ms across the ledger.
GoPostgresKafkaRead the case studyData2025
Atlas Health
Backfills went from 9 hours to 22 minutes.
PythondbtBigQueryRead the case studyPlatform2024
Relay
One slow consumer no longer stalls the other 40.
TypeScriptNATSPostgresRead the case studyProduct2024
Harbor Logistics
Dispatchers handle 3x the daily loads with the same headcount.
Next.jsPostgresRedisRead the case studySecurity2023
Sentinel
Cut the SOC2 evidence pull from 3 weeks to 2 days.
GoTerraformS3Read the case studyProduct2023
Meridian
Search p99 fell below 90ms on a 40M document corpus.
RustOpenSearchRead the case study
04Testimonials
What clients say after we hand over.
“They rebuilt our ledger without a single day of downtime, then stayed on for the six months it took us to trust it. That second part is the rare bit.”
Dana OkaforCTO, Northwind Pay “We had three teams argue about the ingest design for a quarter. CraneDev shipped a working version in nine weeks and the argument stopped.”
Marcus FeldVP Engineering, Atlas Health “The handover was a repo we could actually read, with the reasoning written down. We have added to it for a year without calling them.”
Priya RamanHead of Platform, Relay
05Clients
Teams we’ve shipped for.
- Northwind Pay
- Atlas Health
- Relay
- Harbor Logistics
- Sentinel
- Meridian
06Process
How a project runs.
01Week 1
Read the system
We read the code and the incidents before proposing anything. No discovery theatre.
02Weeks 2–3
Agree the shape
A written architecture with the trade-offs stated, so disagreement happens on paper.
03Weeks 4+
Ship in slices
Something runs in production early and keeps running. No integration phase at the end.
04Ongoing
Hand over, or stay
Your team owns it with the reasoning written down — or we keep the pager. Your call.
Talk to the people who would do the work.
Bring the system you already have. We will tell you where we would start, what we would leave alone, and how long the first release takes.
Book a call