.png)
Can a US Company Hire an ML Engineer in Delhi Without Entity?
Updated: Jul 17
Yes. A US company can hire an ML engineer in Delhi without entity setup by using an Employer of Record that legally employs the engineer in India while the US company directs the work. This avoids Permanent Establishment risk under Indian tax law and typically takes 10 to 15 business days to activate, versus 8 to 12 weeks to register a subsidiary.
We get this question almost every week, and the honest answer is that the without entity part is the easy half of the problem. We've placed machine learning engineers into US companies for a decade from our Delhi desk, and the mistake we see most is founders treating this as a payroll question when it's really a Permanent Establishment question first and a payroll question second.
A US company can hire an ML engineer in Delhi without entity setup by routing the employment relationship through a registered Indian entity, ours or an Employer of Record partner's, that issues the offer letter, runs payroll in rupees, deducts Provident Fund and TDS, and invoices the US company monthly in dollars. The engineer works exclusively for the US team. Nothing else changes except who signs the paperwork.
Why Is Hiring an ML Engineer in Delhi Getting Harder for US Companies?
Delhi NCR is no longer a fallback option to Bangalore for ML talent. It has become a primary market, and that is exactly why sourcing has gotten more competitive for outside companies trying to hire without a local presence.
Gurugram and Noida have absorbed a wave of Global Capability Center builds from BFSI, retail, and healthcare parents recently, and those centers pay aggressively for anyone with production ML experience, not just research background. India today hosts more than 1,800 Global Capability Centers (GCC), employing over 1.9 million professionals and contributing more than 64 billion dollars in value. A meaningful share of the newer ones sit in the NCR belt specifically to be near Delhi's engineering colleges and the Gurugram BFSI corridor.
The knock on effect for a US company without an Indian entity is straightforward. The strongest Delhi ML candidates often have two or three GCC offers on the table before a foreign company even finishes writing a job description. We've watched clients lose candidates simply because a direct hire process, figuring out how to pay someone in India from a US bank account and whether a contractor arrangement is even legal here, took weeks longer than a competing offer from a company already set up through a GCC or an EOR.
The second driver is a widening skills gap. Indian AI talent demand is projected to grow from roughly 600,000 to 650,000 professionals to more than 1.25 million over a five year stretch, per a Deloitte India, meaning demand is outrunning the supply of engineers who can actually ship models to production, not just fine tune notebooks in isolation. That scarcity means Delhi ML engineers with solid MLOps experience are fielding multiple offers within days, not weeks, of going on the market.
Where Do the Best Machine Learning Engineers in Delhi NCR Actually Work?
Delhi NCR's ML depth is concentrated in three clusters, and each brings a different strength and a different gap a US company needs to test for before extending an offer.
Gurugram carries the fintech and BFSI adjacent ML talent, engineers who have built fraud detection, credit scoring, and risk models inside GCCs for large financial services parents.
Noida and Greater Noida hold a younger, product engineering heavy pool coming out of edtech and D2C startups, strong on recommendation systems and NLP but with less exposure to regulated industry model governance. South Delhi and university adjacent talent skews research first, strong math and paper reading ability, weaker on shipping and CI/CD discipline.
What Delhi ML engineers most commonly lack, in our experience, is not technical depth. It is production discipline: model monitoring, drift detection, and writing deployment code that a platform team can actually maintain long after launch. At AnjuSmriti Global, we test for this directly rather than trusting a resume's list of frameworks.
Our screening process has candidates walk through a real incident, a model that degraded silently in production, and explain how they would have caught it. Roughly one in three candidates who list production ML experience cannot answer this cleanly, which tells us they have trained models but never owned one after launch.
Delhi also has a real advantage for US companies specifically: a five and a half to nine hour overlap with US time zones depending on coast, which is workable for standups and code review without requiring anyone to take a call in the middle of the night. This is something Delhi based full stack engineers and ML engineers alike often list as a reason they prefer US based clients over teams with worse overlap.
Can a US Company Hire an ML Engineer in Delhi Without Entity, Legally?
Yes, and the legal mechanism that makes it work is avoiding Permanent Establishment status under India's Income Tax Act, not simply finding a way to send money across borders.
Permanent Establishment is a tax law concept. If a foreign company's activity in India crosses certain thresholds, such as a fixed place of business or an agent habitually concluding contracts on its behalf, the Indian government can treat the US company as having a taxable presence in India, triggering corporate tax filing obligations the company never intended to take on. This is the core legal mechanism that lets a US company hire an ML engineer in Delhi without entity registration in the first place.
An Employer of Record (EOR) sidesteps this because the EOR entity, not the US company, is the legal employer of record in India. The EOR holds the compliance burden: registering the engineer under the Employees' Provident Fund scheme, handling Tax Deducted at Source on salary, and where applicable, complying with Delhi's Shops and Establishments Act on working hours and leave. The US company signs a services agreement with the EOR and a separate assignment letter with the engineer describing scope of work. It never directly employs anyone on Indian soil.
This is also where the difference between contract hiring and full time hiring matters. Contract hiring means engaging someone for a defined project or fixed period, with no ongoing employment relationship once the scope ends, and it works well for a clearly bounded piece of ML work, such as building a single recommendation model over a few months. Full time hiring means an ongoing employment relationship with continuing responsibilities, benefits, and retention expectations, which is what most US companies actually need when they want someone to own a production model long term.
Entity vs EOR vs Contractor: Which Hiring Model Fits a Delhi ML Engineer?
This comparison is also the clearest way to decide whether you actually want to hire an ML engineer in Delhi without entity setup at all, or whether your hiring volume justifies registering one.
Own Indian Entity | Employer of Record | Direct Contractor | |
Setup time | 8 to 12 weeks for subsidiary registration, bank account, and PF or ESI registration | 10 to 15 business days | 2 to 3 days |
Upfront cost | Significant legal and registration fees | None, monthly fee only | None |
Ongoing cost | Full payroll and compliance team, or an outsourced HR outsourcing retainer | 8 to 15 percent of gross salary as EOR fee | Lowest cash cost, highest legal risk |
PE risk | None, the entity is the taxable presence by design | Very low, EOR bears employer status | High if the role is effectively full time |
Best fit | Five or more India hires planned within eighteen months | One to four hires, testing India as a market | Short, clearly scoped project work only |
Benefits, PF, gratuity | Company manages directly | EOR manages, company reimburses | Not applicable, a legal grey zone |
Termination process | Full Indian labour law applies directly to the company | EOR handles notice period and exit compliance | Contractual only, weak protection either way |
If you are hiring one or two ML engineers to validate whether India is the right long term base before committing to global payroll outsourcing or a full subsidiary, the EOR row is almost always the right starting point. Companies that jump straight to paying someone as a contractor because it looks simplest on day one are usually the ones who call us months later needing to unwind a FEMA or Permanent Establishment problem.
How Do We Hire ML Engineers in Delhi for US Companies?
Our process runs on what we call the 72 Hour Shortlist Rule: a shortlist of five to seven vetted candidates within three business days of a completed job brief, because Delhi's strongest ML talent rarely stays unplaced past a week.
Our technical assessment for ML roles runs in three stages: a take home problem built around the client's actual domain rather than a generic exercise, a live pairing session where the candidate debugs a deliberately broken model pipeline, and a systems design conversation covering how they would handle model versioning and rollback in production.
Across our last 32 Delhi ML mandates for US clients, our average time from kickoff to signed offer was 21 days, and we screened an average of 34 candidates for every offer extended, a ratio that tells clients how selective the market actually requires them to be, not just how fast we can fill a seat.
One recent mandate involved a Series B US healthtech company, roughly 40 employees, that needed a senior ML engineer in Delhi to own a clinical risk scoring model, but had never hired outside the US and had no EOR relationship in place. Their first instinct was to hire the candidate as a contractor to keep things simple while they figured out compliance later, exactly the FEMA and PE exposure described above.
We restructured the offer through an EOR before signing, which pushed the start date back by roughly a week and a half. The candidate nearly walked during that delay, choosing a GCC offer instead, but stayed once we could show a committed EOR backed start date in writing. The role closed in 24 days total, and the client's second Delhi hire six months later took just 14 days, because the EOR relationship was already live.
What we would do differently now: on that first mandate, we should have pre cleared the EOR paperwork before shortlisting candidates, not after. AnjuSmriti Global has since moved EOR setup to run in parallel with sourcing on every US to India mandate, which is the single biggest reason our average timeline has come down since.
How Much Does It Cost to Hire an ML Engineer in Delhi Without a US Entity?
Real Delhi NCR machine learning salaries, based on current market data, sit roughly here across three seniority levels.
Mid level, three to five years experience: 18 to 24 lakh rupees per annum
Senior, five to eight years, production ML ownership: 28 to 38 lakh rupees per annum
Lead or principal, eight plus years, team or platform ownership: 42 to 55 lakh rupees per annum
By comparison, a senior ML engineer role in the US market typically carries a base salary in the 150,000 to 220,000 dollar range on public compensation data from sites like Levels.fyi and Glassdoor, meaning the fully loaded Delhi senior hire, even after EOR fees and statutory employer contributions, typically lands at a fraction of the equivalent US total cost.
Total cost via EOR for a senior hire generally breaks down as base salary, plus roughly 13 to 15 percent in statutory employer contributions covering Provident Fund at 12 percent of basic, gratuity provisioning, and group health insurance, plus an EOR management fee of 8 to 15 percent of gross salary depending on volume.
This is the real, all in cost of choosing to hire an ML engineer in Delhi without entity overhead, and clients most often reinvest the savings into a second India hire within the first year rather than banking it, particularly once they see time to value on the first ML engineer.
Conclusion
A US company can hire an ML engineer in Delhi without entity setup today, cleanly and legally, by working through an EOR rather than attempting a direct contractor arrangement. The Permanent Establishment and FEMA exposure of getting this wrong outweighs any short term convenience. In the quarters ahead, we expect this market to tilt further toward engineers who blend classical ML with applied generative AI skills, and in our live mandates right now, US companies that move fast on a committed EOR structure are winning candidates over GCCs that take longer to close.
If you are weighing this decision, our team can walk you through the fastest compliant path: start a conversation with our team.
Interesting Reads:
FAQs
1.Can a US company legally hire a Delhi based ML engineer without registering in India?
Yes. An Employer of Record lets a US company legally engage a Delhi based ML engineer without registering a subsidiary in India. The EOR becomes the legal employer, handling payroll, PF, and TDS, while the US company directs the engineer's actual work day to day.
2.What is the fastest way to onboard an ML engineer in Delhi without a US entity?
The fastest compliant route is an EOR, which typically onboards an ML engineer in 10 to 15 business days once the EOR relationship is active. Setting up the EOR partnership in parallel with candidate sourcing, rather than after an offer, shortens this timeline further.
3.Does Permanent Establishment risk apply if I hire just one ML engineer in Delhi?
Yes, Permanent Establishment risk can apply even with a single hire if the engineer effectively acts as an agent or fixed presence of the US company in India. An EOR removes this risk because it, not the US company, is the legal employer on record.
4.Is it legal to pay a Delhi ML engineer as a contractor from a US bank account?
It can be legally risky if the role functions as ongoing, full time work rather than a short defined project. Under FEMA, recurring salary like remittances disguised as contractor invoices create audit exposure, so genuinely full time roles should use an EOR instead.
5.How much does it cost to hire a senior ML engineer in Delhi through an EOR?
A senior ML engineer in Delhi typically earns 28 to 38 lakh rupees a year, plus roughly 13 to 15 percent in statutory employer contributions and an EOR fee of 8 to 15 percent of gross salary. This still lands well below equivalent US compensation costs.
6.Which Delhi NCR area has the strongest talent for machine learning roles?
Gurugram holds the deepest fintech and BFSI adjacent ML talent through its GCC ecosystem, while Noida and Greater Noida offer a strong product engineering focused pool. The right area depends on whether the role needs regulated industry experience or fast moving product ML skills.
7.Can I convert an EOR hired ML engineer in Delhi to a direct employee later?
Yes. Once a US company registers its own Indian entity, an EOR hired engineer can transition to direct employment with continuity of service preserved and no inherent penalty. Many companies use the EOR phase specifically to validate the hire before committing to full entity setup.
8.What is the difference between hiring an ML engineer on contract versus full time in Delhi?
Contract hiring suits a clearly scoped project with a defined end date and no ongoing obligations. Full time hiring suits an engineer who will own a production model long term, with benefits and retention built in, and is generally best structured through an EOR rather than a contractor agreement.
Comments