Every US company that decides to hire in India starts with the same question and asks it in the wrong order. They ask "should we set up a GCC?" before asking "what work is this team going to own?" — and the second question determines the answer to the first.
This is the decision, in the order it should be made.
Start with the work, not the structure
Three kinds of engineering work go to India, and they need completely different structures.
Execution. Well-specified work, defined elsewhere. A vendor does this well and cheaply, and there is no shame in it.
Ownership. A team owns a service or product area end to end, including its design. This needs your own people, because the context that makes the work good cannot be transferred at the end of a contract.
Leadership. A centre that sets its own roadmap and makes decisions the US team lives with. This needs senior people who could have run something in the US, and they will not join to execute someone else's backlog.
The most common and expensive mistake is pitching ownership work with an execution structure. A senior engineer who could join any Indian product company will not join a US startup's "offshore team" to implement tickets — so you interview for three months, get execution-level candidates, and conclude that Indian engineering talent is weaker than advertised. It is not. Your pitch selected for it.
The four structures
| Start | Cost | Control | Exit | |
|---|---|---|---|---|
| EOR | Days | $300–800/employee/month on top of salary | Medium | Clean |
| Own entity (GCC) | 2–4 months | Setup + compliance overhead, cheapest at scale | Full | Slow |
| Offshore vendor | Weeks | Billed per head, 1.5–2.5× the engineer's cost | Low | Clean |
| Contractors | Days | Highest per hour | Low | Immediate |
EOR is right for almost everyone under fifteen people. You avoid entity setup, statutory compliance and payroll infrastructure, and if the strategy changes you stop cleanly. The economics invert somewhere between fifteen and twenty-five heads, and the softer constraints bite earlier — equity participation is limited, and you are a step removed from your own employment brand.
Your own entity is right once the team is durable and above roughly twenty people. Cheapest per engineer at scale, full control of levelling and equity, and it lets you build an employer brand that compounds.
Vendors are right for genuine execution work, and for testing whether a function can work remotely at all before committing.
Contractors are a bridge, not a structure. Fine for six months; corrosive as a strategy, because nobody accrues context.
The sequencing mistake that costs the most
Do not hire the team before the leader.
The most common failure pattern: a US company hires six good individual contributors in Bengaluru through an EOR, then spends nine months trying to hire a leader who will inherit them. Senior leaders do not want to inherit a team they did not choose, assembled against a bar they did not set. So the search stalls, the six people have no local leadership, and the two best ones leave within a year.
Hire the centre lead first, even though it is the hardest hire and it delays everything. They set the bar, they interview everyone after them, and they are the reason good people say yes — a strong engineer in India joins a person far more often than they join a foreign company's logo.
What it actually costs
Engineer compensation at 2026 rates, converted roughly:
| Level | India (INR) | ≈ USD |
|---|---|---|
| Mid (SDE2) | ₹16–28 L | $19k–34k |
| Senior (SDE3) | ₹28–48 L | $34k–58k |
| Staff | ₹48–85 L | $58k–100k |
| Engineering leader | ₹85 L–2 Cr | $100k–240k |
Then add recruiting (15–25% of first-year salary in India; 8.33% with us), EOR fees or entity and compliance costs, and roughly 42 hours of senior-engineer time per role on interviewing.
The cost nobody budgets is the leadership mis-hire. Getting the first leader wrong typically costs six to nine months and, usually, the team they hired.
Why your search is not working
If you have been interviewing in India for three months and have not hired, it is almost always one of these.
You are screening on titles that do not transfer. An SDE3 at an Indian services company and an SDE3 at a product company can differ by years of real scope. Titles are calibrated internally against the engineers already in that organisation. Assess what someone owned and what they decided alone.
Your pitch describes execution. See above. What the team will own is the single biggest determinant of who applies.
You are only reading applications. The strongest engineers in India are not applying — 72% of the people we place were not looking when we found them. Inbound skews toward people actively in the market, which is a real pool and not the one you need for a founding leadership hire.
Nobody is managing the notice period. Sixty to ninety days is standard at senior level, and counter-offers land during notice, not before it. If nobody stays close to an accepted candidate for three months, you will lose one in four.
What we do
We run India build-outs for US companies — from the first leadership hire down to the scale pod, on one rubric across every hire, then handed over to your own TA team. You do not need an Indian entity or an existing team for us to start, and the same contract covers hiring in the US if you are doing both.
The fee is 8.33% of first-year salary, invoiced on the day the person joins. No retainer, no seat licence, non-exclusive, 90-day replacement guarantee, and no fee at all if an accepted candidate does not join.
We are not the right answer for everything. If you need forty contract engineers next quarter, a staffing firm will be cheaper. If you have not decided what the team will own, no recruiting partner can fix that and we will say so on the first call.
Building or stuck? Send us the role. If the problem is the brief rather than the pipeline, we will tell you that instead — it is the more common diagnosis.