We tell you when the answer is no
Roughly a third of our discovery engagements end with a recommendation to build less than the client asked for. It costs us revenue in the quarter and earns the next three projects.
About
A U.S. consultancy with a global bench
We are structured so that the commercial relationship is simple and local, and the engineering talent is not limited by a commute. One contract, under U.S. law, with senior engineers who overlap your working day.
- Founded
- 2019
- Registered
- United States
- Clients
- Predominantly U.S.
- Network
- 11 time zones
- Contracting
- U.S. law, USD
- IP ownership
- Client, in full
Why we exist
Most failed projects were scoped wrong, not built wrong
We started ML Stack Labs after enough years inside agencies and product teams to notice the same pattern: the code was rarely the problem. The problem was a scope nobody had pushed back on, agreed in a room where saying no would have cost somebody the deal.
So we built the company around being able to say it. Discovery is paid and fixed-price, which means recommending a smaller project does not cost us the engagement. Roughly a third of them end that way. It is the single most useful thing we do.
The second thing is staffing. Every engineer on a client project has cleared our own technical screen and a paid trial task. There is no bench to keep busy, so nobody is placed on work they are wrong for to hit a utilization number.
We are U.S.-registered and contract under U.S. law, with commercial and engagement leads in Eastern Time. Delivery draws on an engineering network spread across eleven time zones, with at least four hours of daily overlap guaranteed on every placement.
Senior people, on the actual work
The engineers in the scoping call are the engineers who write the code. There is no bench, no bait-and-switch and no delivery layer you meet after signing.
Your repositories, your cloud, your keys
We commit into your version control and deploy into your accounts from day one. Nothing needs migrating when the engagement ends, because nothing ever lived with us.
Handover is the deliverable
An engagement has gone well when your team ships the next change without calling us. Runbooks, decision records and paired on-call are scoped in, not bolted on.
- 60+
- Products shipped since 2019
- 94%
- Client retention past three years
- 11
- Time zones engineering network
- 12 days
- Median time to staff from signed SOW
Companies we have delivered for
Sully AI
Pony.ai
Meridian Health
Northlake Capital
Vantage Pay
Alder Diagnostics
Brightpath Learning
Cardinal Bancorp
Tillage Systems
Axle Motion
Halcyon Care
Ledgerline
Summit Scholars
Rowan Robotics
Sully AI
Pony.ai
Meridian Health
Northlake Capital
Vantage Pay
Alder Diagnostics
Brightpath Learning
Cardinal Bancorp
Tillage Systems
Axle Motion
Halcyon Care
Ledgerline
Summit Scholars
Rowan Robotics
How we work
Four stages, and you can stop after any of them
No twelve-month master agreement before anything is built. Each stage produces something you own outright, whether or not you continue.
Around a third of discovery engagements end with us recommending a smaller project than the client came in asking for. We would rather find that in week two than month six.
- 0145 minutes
Scoping call
You walk us through the problem and the constraints: budget, deadline, and the internal politics we should know about. We tell you honestly whether it is something we should take on.
- Written summary
- Rough size and shape
- Go / no-go
- 022 weeks
Discovery
Paid and fixed-price. We read the code, talk to whoever maintains it, and produce an architecture, a risk register and a sequenced roadmap you own outright.
- Architecture doc
- Risk register
- Costed roadmap
- 032-week increments
Build
A dedicated pod works in your repositories with a demo every second Friday. Scope is reviewed at each increment boundary, which is also where you can stop.
- Working software
- Demo each increment
- Weekly written status
- 042 to 4 weeks
Handover
Runbooks, architecture decision records and paired on-call shifts until your team is running the system without us. We stay reachable afterwards on a support retainer if you want one.
- Runbooks and ADRs
- Paired on-call
- Optional retainer
In their words
What clients say when the project is over
They spent the first two weeks telling us the project we asked for was the wrong one, and were right. The version we actually built was smaller, cheaper and is still running.
The ledger cutover was the least dramatic infrastructure change we have ever done. Twenty-one days of parallel running and then it was just done.
We had been quoted a rewrite by two other firms. ML Stack Labs carved it up and shipped the first piece in five weeks without taking the system down.
Tell us what is on fire.
Bring us a stalled project, a system nobody wants to touch, or a product you need built properly the first time. We will tell you honestly whether we are the right team for it.
- 45 minutes, no cost, no deck
- You speak to an engineer, not a salesperson
- We say no if it is not a fit