Five ways to buy · one bench of nine
PHP Development Services
Laravel, CodeIgniter and legacy PHP engineering from Pune, for product teams and agencies in the US, UK and Canada. The same nine salaried engineers sit behind all five services below, what changes is who carries the risk, and what you pay for that.
3 engineers free now · all nine within four weeks
The five services
Every way we sell PHP engineering
Same engineers, same review standards, same published rate. What differs is who owns the scope and who carries the schedule risk.
PHP 8.4APIsLegacy rescue Best when scope keeps moving Hire Laravel developers STAFF AUG Laravel 12, Livewire and Filament in production, plus the queue and tenancy work that decides whether a SaaS survives its own growth. 
Laravel 12HorizonFilament Best for product teams Hire CodeIgniter developers AUDIT FIRST A two-day audit that answers one question in writing: patch it, upgrade it, or migrate it. Then whichever of the three is honest. CI3 & CI4PHP 8.4Rector Best for inherited apps PHP development company FIXED SCOPE One price, one date, a late penalty and a defect warranty. We own scope, QA and delivery; you approve a spec. Best for a hard deadline PHP outsourcing for agencies WHITE-LABEL Capacity under your name, in your Slack, on your NDA. Your client never learns we exist, and the non-solicit runs both ways. Best for agency pipelines Capabilities
What the work actually is
Roughly half of what we do arrives on a stack nobody would choose today. That is the speciality, not the compromise.
Laravel application development
SaaS platforms, multi-tenant apps and the internal tools that replace a spreadsheet nobody trusts.
Laravel 12
Livewire 3Filament 4 Legacy PHP rescue and upgrade
PHP 5.6 to 8.4 without a rewrite: Rector and static analysis first, then a regression suite around the money paths.
Framework migrations
CodeIgniter or Symfony to Laravel, module by module behind one router. Both frameworks live until the last one moves.
CodeIgniterStranglerZero-downtime API and integration work
Payment gateways, ERPs, PIMs and shipping carriers, idempotent, replayable, and monitored rather than hoped over.
Stripe Queue and performance work
Horizon topologies, Octane, Redis and the query work that turns a four-second page into a fast one.
Redis Test suites and CI
From no tests to a coverage floor that blocks a merge, with deploys automated out of your main branch.
GitHub ActionsStaging Straight answer
Why PHP, in 2026
Nobody asks whether PHP is fashionable. They ask whether the thing already running their business can keep running, and whether the next feature will take two weeks or two months. That is a maintenance question, not a language question, and it is most of our work.
PHP 8.4 is a genuinely fast, typed, well-tooled runtime, and Laravel is as productive as anything in the Node or Python ecosystems for the applications most businesses actually build. The problem is rarely the language. It is nine years of untested code written by four people who have left.
So we will not tell you to rewrite in something modern. We will tell you what the upgrade path costs, what the rewrite costs, and which one we would choose with your budget, including the times when the honest answer is to leave it alone for another year.
We have talked six enquiries out of a rebuild this year. Two of them came back with work that genuinely needed doing.
In every engagement
Six terms that do not change
- IP assigned on the first commit
- Not on final payment. An assignment clause, not a licence.
- Ten working days risk-free
- Real tickets. Walk away at the end and owe nothing.
- NDA before repository access
- Yours or ours, signed before anyone joins your organisation.
- Published rates, one price for all
- No regional pricing, and no discount for negotiating hardest.
- Thirty days’ notice to stop
- Month to month after the trial. No annual minimum, no exit fee.
- Named engineers, not a pool
- The same people for the engagement, named in the contract.
Proof, not adjectives
What we hold every commit to
Not a portfolio. These are configured in CI, so a pull request that misses them cannot merge, and you can read the configuration in your own repository during the trial.
Where the work comes from
The four situations that bring people to this page
Eight years of enquiries fall into a small number of shapes. Recognising yours saves a call.
One person knew the system, and now nobody does
Usually a Laravel or CodeIgniter application built by a contractor or an early employee, with no tests, no documentation, and a deploy process that lives in somebody’s head. The business runs on it. Our first month here is spent writing down what exists, because you cannot safely change software nobody can describe. This is the single most common reason people call us.
The host is dropping the PHP version you run on
A hard external deadline, usually eight to twelve weeks out, and a codebase that will not start on anything newer. Rector and static analysis do most of the mechanical work, but the interesting part is the code that runs without erroring and produces subtly different results. That is why we write a regression suite around the money paths first.
It worked at a thousand users and does not at fifty thousand
Almost always the database rather than PHP: a missing index, an N+1 inside a loop, a nightly import competing with customer requests. We measure before touching anything, because the component everybody blames is usually fine and the real cause is boring. Occasionally the answer is a queue and a cache rather than a rewrite.
An agency finished, and now you own something you cannot read
This starts with a paid audit rather than a quote, because we need to know what is actually there before putting a date against it. The report tells you what is solid, what is risky, and what the previous team left half-done. You own it whether or not you continue with us.
How we estimate
Why our first answer is usually a question
An estimate is a claim about work nobody has done yet, so the honest version comes with its assumptions attached. When we send a number, it names what we assumed about your codebase, and it names the two or three things that would change it. If you have ever received a single figure with no reasoning, you were being sold certainty rather than an estimate.
For anything substantial we would rather run two paid weeks of discovery than guess. That produces a written specification, an architecture, a risk list and one fixed build price instead of a range. It costs a fixed fee, it belongs to you afterwards, and several clients have taken ours to other agencies and used it to get better quotes. We think that is fine.
For smaller work we quote on the spot, usually within a day, because the range of outcomes is narrow enough that a buffer covers it. A version upgrade on an application we have read, a single integration, a defined bug: none of those needs two weeks of process wrapped around it, and adding process to small work is how agencies justify their own overhead.
The thing that moves an estimate most is not technical. It is whether somebody on your side can answer a question the same day. An engineer waiting three days for a decision is an engineer you are paying to be blocked, and it is the most expensive avoidable cost in this kind of arrangement.
We have never sent a fixed build price without discovery, and we are not going to start. A firm number from a two page brief is either padded heavily or planning to make the difference back on change requests.
What we will tell you no about
Four things we turn down
- A rewrite you do not need
- If the existing application works and earns money, the upgrade path is usually a third of the cost of replacing it. We have talked six enquiries out of a rebuild this year.
- Fixed price on a moving scope
- Fixed price needs a fixed specification. Without one you are buying a buffer, and you would do better hiring a developer by the month and directing the work yourself.
- Work with no reviewer on your side
- If nobody with product context can answer questions, no arrangement survives it, ours included. We would rather say that in week one than bill you through it.
- A stack we do not know
- We write PHP, JavaScript and SQL. If your problem is really a data engineering or mobile problem, we will say so and point you somewhere better.
The stack, in detail
What we actually run, and why
Not a logo wall. These are the choices we make by default and the reasoning behind each, so you can disagree before we start.
Framework and language
Laravel 12 for anything new, because the ecosystem around queues, testing and deployment saves more time than a lighter framework saves in overhead. CodeIgniter 3 and 4 we maintain and migrate rather than recommend. Plain PHP applications we treat with respect: a well organised 2015 codebase is often easier to work in than a badly organised modern one.
PHP 8.4Laravel 12CodeIgniter 3 & 4
Symfony components Data and caching
PostgreSQL where we get a choice, MySQL 8 where we do not, and Redis for cache and queues on anything with real throughput. Database queues are fine to start and become the bottleneck sooner than people expect. Most performance problems we are called about are query problems, so we start with slow query logs rather than adding infrastructure.
PostgreSQL 16
MySQL 8RedisHorizon Testing and static analysis
Pest for new suites, PHPUnit where one already exists, PHPStan at level 8 on new code and level 5 minimum on anything inherited. Coverage on business logic is a merge gate rather than a target, and the paths that touch money get covered first. This is the part clients are most surprised is included rather than billed.
Laravel PintGitHub Actions Hosting and deployment
Your infrastructure, in your account, set up in your name. AWS, Hetzner and Laravel Forge are what we use most. Deploys are automated from your main branch through a staging environment, never from a laptop, and every project gets a runbook so the person on call at three in the morning is not reading source code.
HetznerForgeStaging first FAQ
Questions before the first call
If yours is not here, email it. We publish the answer rather than saving it for a call.
What does a PHP development company in India actually cost?
Hourly rates start at $22 and every band is published on the rates page, no gate, no call required. Fixed-scope builds start at $9,500 after a paid discovery, and a two-day audit is $800.
Which of these five services do I need?
If scope is moving, hire a developer. If the deadline is external, buy fixed scope. If you inherited the code, start with the audit. If you are an agency, take a retainer. We will say which on the first call, including when the answer is none of them.
Do you work with PHP other than Laravel?
Yes, CodeIgniter 3 and 4, Symfony components, and plain PHP applications that predate all of it. Roughly half our work arrives on something nobody would choose today.
How do you handle a codebase with no tests?
We write a regression suite around the paths that touch money before changing anything, then work outward. Coverage on business logic is a merge gate, not an aspiration.
Can you take over from another agency?
Yes, and it starts with a paid audit rather than a quote. We need to read what is actually there before putting a date against finishing it, and you own that report regardless of what you do next.
What are your working hours against ours?
Four hours of daily overlap with EST as standard: standup at 9:00am your time, and a written handover in your inbox before your morning.
Who owns the code?
You do, assigned from the first commit rather than on final payment. Work happens in your repository, so there is no handover event to negotiate.
How quickly can someone start?
Three engineers are free now and the rest come free within four weeks. A shortlist with rates and start dates reaches you inside one business day.
Do you work with startups or only established businesses?
Both, though the honest answer differs. An established business with a working application and a real deadline is our best fit. A pre-revenue startup usually needs to move faster and cheaper than a fixed-scope process allows, so we would suggest hiring one developer by the month and directing them yourself.
What if we already have developers?
That is a common arrangement and it works well. Our engineer joins your standup, follows your conventions and reviews alongside your team. What we ask is that somebody on your side has review capacity, because a pull request sitting unread for four days is the most expensive thing in this arrangement.
How do you handle security and access?
NDA before repository access, least-privilege credentials, no production database copies on laptops, and access revoked the day an engagement ends. If you have a vendor security questionnaire, send it, and we will fill it in rather than ask to be exempted.
Can you sign our client’s NDA as well as ours?
Yes, and we have done it repeatedly. Agencies frequently need our agreement to flow down to whatever their client requires, and we sign those without argument as long as the terms are not mutually contradictory.
Do you take equity or work for a reduced rate?
No. We have been asked and we always decline, because it changes the relationship from a supplier who can be fired into a partner who cannot. The published rate is the rate, and it is the same for everybody.
How quickly can you start on something urgent?
Three engineers are free now, so a defined piece of work can usually begin within a week of the first email. Anything needing discovery adds two weeks before build starts. If we cannot help inside your window we will say so on the first reply rather than string it out.
Do you provide written status reporting?
Weekly, in writing, whether or not you ask. It covers what shipped, what is in progress, what is blocked and what we need from you. Pod engagements also get sprint reports and a burn-down. There is no separate project management fee for any of it.