Senior AI & software engineering

For the AI projects your team can't crack.

GPU inference clusters. Autonomous calling agents. Vision AI in live logistics operations. HIPAA-compliant healthcare systems. We've built them at Y Combinator startups and our own company. Bring us the hard one, and work directly with the engineer who built those systems.

Free 30-minute technical call. You'll leave with a clear plan, whether or not we work together.

Built AI systems at Reacher YC S25 Careforce AI YC F24 Attensive Logistics AI Shifthive Mount Sinai NYC ShiftPosts Acquired
What we're hired for

Problems where getting it wrong is expensive.

We take a small number of engagements, each one a hard problem with real stakes. If a generalist agency could do it cheaply, we'll tell you.

Typical starting points

  • "Our AI prototype works in demos but fails on real data."
  • "We have video or camera data and no way to turn it into answers."
  • "We need AI agents handling calls or documents at volume, and they can't get it wrong."
  • "Enterprise customers want security answers we don't have."
  • "Our inference costs are out of control."
  • "We raised money and need a technical leader before we can hire one."
Fixed price · 1–2 weeks

Not ready for a full build? Start with an AI Production Audit.

We measure your AI system's accuracy on real data, reliability, cost, and security, then hand you a prioritized fix plan.

See the audit →
Track record

We've already solved the hard parts.

These aren't demos. They're production systems with real users, real load, and real consequences when they fail.

Millionsof TikTok videos indexed by a self-hosted AI pipeline
< 2 secsemantic search across the full video index
80–90%GPU utilization on an 8× H100 inference cluster
90%of a YC healthcare startup's production code, built and led
Selected work

From GPU clusters to phone calls.

Reacher · YC S25

AI video search at scale

A distributed pipeline that watches, transcribes, and understands millions of TikTok videos, then makes them searchable in under two seconds.

Careforce AI · YC F24

Autonomous AI calling agents

Voice agents that call patients, understand the outcome, book appointments, and schedule follow-ups across multiple health centers.

Attensive

Vision AI for logistics

Our own product. It turns existing facility cameras and phones into live measurements for parcels, trailers, and inventory.

How we work

A clear path from idea to launch.

Scope

A free call to understand the problem, the users, and what success looks like. You get a written plan and estimate.

Prototype

We build the riskiest piece first, so you see something real within the first week or two.

Build

Weekly demos, a shared board, and direct access to the engineer. No surprises at the end.

Launch & support

Deployment, monitoring, documentation, and handover. Or we stay on to keep improving it.

Why Demist

You work with the person who built those systems.

Most agencies sell you a senior engineer on the call and hand your project to a junior team. At Demist, the engineer who built AI infrastructure at two YC startups is the one who scopes, architects, and builds your project.

  • Founder mindset. We've started and sold companies, so we build for business outcomes, not billable hours.
  • Production-grade from day one. Monitoring, error handling, fallbacks, and security are part of the build, not an add-on.
  • Full-stack AI depth. From prompt design to GPU clusters, from hosted APIs to self-hosted open-source models.
  • Regulated industries. HIPAA-compliant infrastructure, security questionnaires, and enterprise contract reviews.
  • You own everything. Your code, your accounts, your data. Fully documented.
Makers of Attensive

We build our own AI product, too.

Demist AI is the company behind Attensive, AI vision for logistics. It turns existing facility cameras and smartphones into live measurements for parcels, trailers, and inventory, and it's in pilots with large enterprise logistics operators.

  • Every client project benefits from what we learn shipping our own AI to demanding enterprise customers.
  • We know what it takes to keep AI accurate outside the lab: real cameras, real lighting, real operators.
  • Visit Attensive →
Engagements

Start small, or go all in.

Start here

Discovery Sprint

1–2 weeks

  • Requirements & technical plan
  • Architecture and cost estimate
  • Working prototype of the riskiest part
Start a sprint
Ongoing

Monthly Retainer

Month to month

  • Continued development & support
  • Fractional CTO leadership
  • Priority response
Discuss a retainer
FAQ

Common questions

How much does a project cost?

It depends on scope. After a free scoping call, we send a written proposal with a fixed price or a monthly rate, so you know the cost before anything starts. Many clients begin with a short Discovery Sprint to lower risk.

How fast can you start?

Usually within one to two weeks of a signed proposal. You'll see a working prototype early in the engagement.

Do I own the code?

Yes. All code, infrastructure, and accounts belong to you. We work in your repositories and cloud accounts whenever possible.

Can you work with our existing team?

Yes. We can embed with your in-house team, review architecture, and help hire and onboard engineers, as we did when building the engineering team at Careforce AI.

Do you sign NDAs?

Yes, happily, before you share anything sensitive.

Have a hard problem? Let's solve it.

Tell us what you're building. We'll reply within one business day.

Book a free consult →