That posting readslike one system.
Not a seat.
Sometimes a build role really is a build role and you should go and hire for it. The other possibility is that what sits underneath the posting is one system that needs to exist, and a system is a scoped piece of work rather than a change to the org chart.
That second version is the one we do. Scrapers, enrichment pipelines, MCP servers so your agents can call any of it directly, and the plumbing that keeps the data moving once we've handed it over.
01 / Scope against headcount
The same build, two ways of paying for it.
The posting might be exactly right. Whether it's a seat or a system is the only thing on this page we're arguing with.
The posting
The project
- Getting started
The posting
Two months of interviews, a take home, references, then a notice period.
The project
You send the spec. We tell you what it takes to build it.
- What you commit to
The posting
A senior salary and a seat on the org chart, decided before the work is scoped.
The project
One system, scoped, and it ends when the system is built.
- If it's the wrong shape
The posting
You find that out two quarters in.
The project
You find that out from the first build, and the work is yours either way.
02 / What gets built
What that system is made of.
- Scrapers
- Sites, boards, registries and directories, with the retry and proxy behavior that keeps them running after week one.
- Enrichment pipelines
- Waterfall order, cost per row, and a confidence field on every match, so a weak one is visible instead of silent.
- MCP servers
- Every tool exposed so an agent calls it directly, with typed inputs and flat output at one row per domain.
- The plumbing
- Schedules, alerting, and the schema that holds when the shape of the source data changes on you.
03 / The fleet
Judge the work before you spend twenty minutes on a call.
57 of our actors are live on the Apify Store under Mamba Labs, with 722 people running them. You don't have to take our word for any of it. Open one, read the input schema, run it against a domain you already know, and see what comes back.
They cover identity resolution, firmographics, hiring and funding signals, tech stack detection and ICP scoring. Every one returns flat JSON at one row per domain, and every one has an MCP server so an agent can run it without a human in the middle.
Eleven years of software engineering at Facebook, IBM, Instacart and Dolby sat behind those before we built the first one.
- Public actors
- 57
- People running them
- 722
- Counted live on
- 2026-10-16
04 / Handover
Documented while it's built, not after.
The docs go up alongside the code rather than in the last week of the engagement. What got built, why it got built that way, and the rules underneath it.
Everything we build for you is yours. The code, the schemas, the docs, the data. Whoever picks it up after us reads how it works instead of guessing at it.
05 / Not for
Three reasons this isn't a good fit
- The work is continuous, and it's the same work every week
- That's a seat. Fill it.
- You need someone owning a roadmap and a team
- Also a seat. We build one system at a time.
- You want a duration before you've sent a spec
- We won't quote a timeline against a scope we haven't seen. Send the spec and you get a real answer instead of a comfortable one.
06 / The ask
Send the spec.
Whatever you're trying to build, written however you've got it: a ticket, a paragraph, the job description itself. We'll read it and tell you what it takes to get it done, and whether it's a project or genuinely a hire.
Booking for something else? Book anyway and say so, or use cal.com/mamba/quick-intro for the general calendar.
Rather write it than book a call?
Then write it. It reaches us directly and we answer within a day.