Your accounts, not ours
Cloud infrastructure is created in your account and billed to you, and the repository is yours from the first commit. We work as collaborators with access that can be revoked the day a project ends.
How we work
How a project runs, what we need from you, how your data and credentials are handled, what happens after launch, and where the commercial terms sit. If you are comparing vendors, this is the page to send to the person who has to sign.
Delivery
The right-hand column matters more than the left. Most delays we have seen were not engineering problems.
We look at the problem, the systems already in place and the constraints nobody mentions in a brief — the air-gapped network, the ERP that cannot change, the regulator. You leave with an architecture, a scope and a fixed price. This stage is paid and it is the one where we are most likely to tell you not to build the thing.
What we need
An hour with whoever owns the process, and access to one real example of the data.
Flows and screens for anything people touch; schema, API shape and infrastructure for everything behind them. The milestone schedule and payment plan are fixed here, and the scope document becomes the thing we both point at when something changes.
What we need
Feedback within three working days of each review. Timelines assume that and slip without it.
Work lands on a staging environment you can open at any time — not a demo at the end. Source is in your repository from the first commit, infrastructure in your cloud account. You see the thing weekly, and we would rather hear a problem in week three than in week ten.
What we need
A named decision-maker who can answer a question in a day.
Access control, error handling, logging, backups, load behaviour and the unglamorous edge cases. Then deployment, a runbook, and training for the people who will use it. Hardware adds a pilot in the field here, because only the field finds some problems.
What we need
Whoever will operate it, for a training session that is genuinely attended.
Thirty days of bug-fix warranty from delivery. Then either a maintenance retainer, or a clean handover with the architecture notes, runbook and an overlap period with the team you hire. Both are real options and we are not offended by the second.
What we need
A decision on which, before the warranty ends.
Security and data handling
Written plainly, including the certifications we do not hold — because the row you are really scanning for is the one a vendor left out.
Cloud infrastructure is created in your account and billed to you, and the repository is yours from the first commit. We work as collaborators with access that can be revoked the day a project ends.
We ask for the least access that lets the work happen, through your own secret manager or a shared vault — never over email or chat. Keys live in environment configuration on the server, never in frontend code or a repository.
Where we need real data to build against, we ask for a de-identified extract. Where that is impossible we agree in writing what may be copied, to which machines, and when it is deleted. Production data is not copied to a laptop as a convenience.
Encryption in transit and at rest, least-privilege access, audit logging on reads of sensitive records, documented retention and deletion. For anything clinical, financial or personal these are specified in week one rather than retrofitted.
Where a project uses a model provider we use the enterprise or API tier under which the provider does not train on your inputs, and for sensitive work we run open-weight models on infrastructure you control. Which of the two applies is written down before the build starts.
We are not SOC 2 or ISO 27001 certified, and we will not imply it. We build to the practices above and will complete your security questionnaire honestly, including the rows where the answer is no.
After launch
A system nobody maintains is a liability with a login page. Here is what happens in month two and in year two.
30 days from delivery, included. A bug is behaviour that differs from the agreed scope; a new feature is a new feature, and we will say which one a request is.
Production down: same working day. Something broken but working around: two working days. Everything else: the next change window.
A published queue you prioritise, so small changes do not each become a negotiation.
Dependency upgrades, security patches, certificate renewals, cloud cost review. The work that is invisible until it is not done.
Thirty days' notice, a handover document and an overlap call with whoever takes it on. No exit fee, and no hostage-taking of the deployment.
Procurement & commercial
Ownership, pricing, liability, who does the work and what happens when it goes wrong — answered here rather than in a call you have to book first.
You do, on final payment for each milestone — source, designs, schema and documentation. We keep the right to describe the work in general terms for our own portfolio, and even that stops if you ask. Anything we built before your project, or generic utilities, stays ours and comes to you under a perpetual licence so nothing in your system depends on our goodwill.
Startups, enterprises and product teams — from a napkin sketch to a fleet in the field. One team, concept to production.
Start a project
Send us the constraint that worries you most. An engineer reads it — not a form queue — and replies within one working day.
Lucknow, Uttar Pradesh · India