Salary benchmarks for startups in India and the US

1. The core idea#

A salary band is a decision you make once and then defend for every hire. If you guess at each offer, you get three problems at once: you overpay the confident negotiator, you underpay the quiet scientist who later finds out, and you burn runway faster than the plan assumed. A benchmark is not the answer to "what should we pay"; it is the reference point you deliberately sit above or below, with equity making up the difference (5.12).

Who this is for: a deeptech founder making the first 5–25 hires, usually split between India and the US (sometimes the UK), who needs defensible bands before the first offer goes out.

The three rules this playbook works from:

• Pick a percentile on purpose. Most seed companies pay cash at the 25th–50th percentile of funded-startup data and make up the gap in equity. Decide this once, write it down, and apply it to every role.
• Benchmark against the right population. Big-tech and IT-services data are the wrong comparison for a seed startup. In India in particular, the spread between an earlystage startup and an MNC product company for the same title is often 3–5x, so a single "market rate" is meaningless.
• Compare total cost, not headline salary. An Indian CTC already includes employer PF and gratuity; a US salary does not include payroll tax, health insurance or 401(k) match. Budget on loaded cost (section 4).

Related: 5.12 (cash vs equity trade-offs), 5.14 (paying PhDs and scientists), 5.17 (bonuses and sales comp), 5.02–5.03 (ESOP pool sizing).

2. What the data says (November 2026)#

Salary data is noisy. The sources below differ in sample, method and definition (base vs total cash, CTC vs in-hand), so use them to set a range and a direction, not to justify a single number. Where a figure comes from an aggregator rather than a payroll dataset, it is marked as such.

United States#

Two things stand out. First, early US startups are now competing for engineers with equity more than cash: grant sizes at small companies have grown several times faster than salaries. Second, the spread is wide. A senior founding engineer at the 50th percentile earns close to what a mid-level engineer earns at Series C, because seniority and scarcity matter more than stage at the very first hires.

India#

There is no Indian equivalent of Carta's payroll dataset with wide coverage. The best public sources are consultancy reports and aggregators built on self-reported data (AmbitionBox, Levels.fyi, PayScale). They broadly agree on the following:

The pattern in India is a steep split by employer type. The same title can pay ₹8 LPA at a seed startup and ₹40 LPA at an MNC product company. AI/ML, data and specialised hardware or embedded roles carry the largest and fastest-growing premium. Notice periods of one to three months are normal and shape hiring timelines as much as pay does.

United Kingdom (for comparison)#

Ravio's 2026 data shows a mid-level software engineer median of £70,800 at preseed/seed, £73,900 at Series A/B and £76,100 at Series C+. That is only a 7% spread at mid-level, much smaller than in the US or India. Senior engineering managers show a bigger gap: £108,400 at early stage against £129,200 at growth stage. Mid-level engineer pay at early-stage companies rose 12.6% in 2026, faster than at later stages. Median variable pay at early-stage companies is 11.5% of base.

What this means for a deeptech founder#

• For the first few hires, seniority and scarcity set the price, not your stage. A senior hardware or ML engineer will benchmark against later-stage offers whatever round you have raised.
• Equity is doing more of the work in 2026. If your cash is at the 25th percentile, your grant should be clearly above the median for your stage (5.02, 5.12).
• India data is the weakest. Triangulate at least three sources and check against offers your candidates actually hold.

3. How to do it: building salary bands#

Step 1. Write a one-paragraph compensation policy#

Decide three things and write them down before the first offer:

• Cash percentile: for example, "we pay base salary at the 40th percentile of fundedstartup data for the role and city".
• Equity percentile: for example, "we grant equity at the 60th–75th percentile for our stage" (5.02, 5.12).
• Geography: one band per city tier, or one global band. Most India–US deeptech companies use separate country bands, with India split into metro and non-metro.

This is the document you show a candidate, a new hire who compares notes, or an investor who asks how you set pay.

Step 2. Define 4–6 levels, not titles#

Use a simple ladder: for example, L1 junior, L2 engineer, L3 senior, L4 staff or lead, L5 principal or head of. Write one line on what separates each level (scope, autonomy, who they lead). Benchmark by level, not by title, because startups inflate titles and candidates know it.

Step 3. Triangulate three sources per role#

For each level and country, collect at least three data points:

1. A dataset: Carta Total Comp, Pave, Ravio (UK and Europe) if you have access through a platform you already pay for. In India, consultancy reports (xto10x, Aon, Mercer) or a recruiter's internal data.
2. Self-reported data: Levels.fyi, Glassdoor, AmbitionBox, Wellfound. Filter to funded startups where you can.
3. Live market evidence: the current pay and competing offers of candidates you have actually interviewed. This is the most accurate source, and it goes out of date fastest.

Take the 25th, 50th and 75th percentile from the overlap. If the sources disagree by more than 30%, the role is probably defined differently in each; narrow the definition.

Step 4. Set the band#

Put the band midpoint at your policy percentile. Make the band about ±15% around the midpoint (±20% for senior roles). Offers should normally land between the band minimum and the midpoint, leaving room for raises without a promotion.

Step 5. Calculate the loaded cost and check it against the plan#

Add the statutory and benefit costs (section 4) and check that the hiring plan still fits the runway (16.03 Budgeting and burn). Do this before you make the offer, not after.

Step 6. Review the bands every 12 months#

In 2026 mid-level salaries are moving 5–13% a year in some markets (Carta, Ravio). Bands that are 18 months old will lose you candidates. Review them at each fundraise and at least once a year.

Worked example#

A battery-materials startup with its lab in Bengaluru and a commercial lead in the San Francisco Bay Area has raised a $3M seed round. Policy: base salary at the 40th percentile, equity at the 70th. Currency figures are illustrations at an assumed ₹ 88 to $1.

Role A: senior embedded engineer, Bengaluru (L3, 6 years' experience)

The 40th percentile is about ₹32 LPA, so the band is ₹27–37 LPA. The founder offers ₹33 LPA fixed CTC plus an above-median ESOP grant (5.03), and explains the band to the candidate.

Loaded cost: the CTC already includes employer PF. Add gratuity provision (4.81% of basic; basic is 50% of wages under the Labour Codes, so about ₹16.5 lakh × 4.81% ≈ ₹0.8 lakh) and group health insurance (about ₹0.25 lakh). Total about ₹34 lakh, or about $39,000 a year.

Role B: founding commercial engineer, San Francisco Bay Area (L3)

Pave's senior founding-engineer benchmark is $187k at the median. The 40th percentile is about $180k, so the band is about $153k–$207k. The founder offers $180k.

Loaded cost: employer Social Security of 6.2% (the 2026 wage base is $184,500, so the full $180k is taxed) is $11,160. Medicare at 1.45% is $2,610. Federal and state unemployment tax is about $1,000. Health insurance and a modest 401(k) match come to about $20,000. Total about $215,000 a year, roughly 1.2x base.

What the founder learns

Role B costs about 5.5 times Role A. With $3M of seed money, that difference decides whether the company can afford a third lab engineer. The founder keeps the lab in Bengaluru and hires one US person, the role that has to be face-to-face with customers. This is the same decision as 16.03–16.04 (runway) and 5.18 (equity for overseas staff): set the bands first, then the hiring plan.

4. Country notes: what sits on top of the salary, and pay transparency rules#

India#

• Quote and compare CTC, but budget on loaded cost. Indian offers are quoted as cost to company (CTC), which normally already includes employer provident fund (PF). Inhand pay is typically 55–75% of CTC, depending on the tax regime and deductions. Candidates compare in-hand pay, so show both in the offer letter (5.11).
• The Labour Codes changed salary structure. The four Labour Codes came into force on 21 November 2025. Under the new definition of "wages", basic pay plus dearness allowance and retaining allowance must be at least 50% of total pay. Excluded allowances above 50% are added back. This raises PF and gratuity costs compared with the old habit of a low basic and large allowances. Re-check any salary template written before November 2025.
• PF: employer contribution of 12% of basic wages. Many startups cap it at the ₹15,000 a month wage ceiling; contributing on the full basic is more generous and more expensive. State the choice in the offer.
• Gratuity: 15 days' wages for each year of service, payable after five years (one year for fixed-term employees under the Code on Social Security). Provide for it at about 4.81% of basic.
• Other costs: group health insurance (often ₹15,000–40,000 a year per employee for a family policy), notice-period buyouts (often ₹1–3 lakh) and recruiter fees.
• Tax regime: most employees now opt for the new regime under the Income-tax Act 2025, which removes most allowance-based tax savings. Flexible benefit plans (food, fuel and so on) matter less than they used to.

United States#

• Payroll taxes: employer Social Security of 6.2% up to the 2026 wage base of $184,500, plus Medicare of 1.45% with no cap, plus federal and state unemployment tax. For most startup salaries this adds about 8–9% to base.
• Benefits: health insurance is the big cost. Bureau of Labor Statistics data put total benefits at about 30% of total compensation across private employers. A common startup budget is 1.2–1.3x base salary for fully loaded cost.
• Exempt status: to treat a salaried employee as exempt from overtime, they must pass the duties test and earn at least the federal minimum ($684 a week after the 2024 rule raising it was vacated). Several states set higher floors; California's is twice the state minimum wage for full-time work. Check this for junior lab and operations roles.
• Pay transparency: a growing list of states require a good-faith salary range in job postings, including California, Colorado, Washington, New York, Illinois, Maryland, Minnesota, Hawaii, Massachusetts, New Jersey and Vermont. Virginia (1 July 2026), Maine (29 July 2026, employers with 10 or more staff) and Connecticut (1 October 2026) were added this year. Many states and cities also ban salary-history questions. If you hire remotely, assume the strictest rule applies.

United Kingdom#

• Employer National Insurance: 15% on earnings above £5,000 a year (since April 2025). The Employment Allowance can offset up to £10,500 a year for eligible employers.
• Pension: auto-enrolment requires at least 3% employer contribution on qualifying earnings.
• Pay ranges: not yet a legal requirement in the UK, but more than half of UK job postings already show salary information (Indeed Hiring Lab, March 2026), and candidates expect it.

European Union#

• The EU Pay Transparency Directive had a transposition deadline of 7 June 2026. By late September 2026 only five of 27 member states had transposed it in full (Italy, Malta, Lithuania, Slovakia and Greece), with partial rules in force in others such as Poland and Estonia.
• Where it applies, employers must: give candidates the starting pay or range before interview, not ask about salary history, and answer employees' requests for comparator pay information. The candidate rules apply to employers of every size.
• For a deeptech startup hiring a European scientist: publish a range and never ask what they currently earn, even if the country has not yet passed its national law.

5. Scripts, mistakes and sources#

Scripts#

Explaining your band to a candidate

"We set pay the same way for everyone. For this level in Bengaluru our salary band is ₹27–37 lakh, and we put base salary at around the 40th percentile of funded startups. We pay above the median in equity because we think that's where the upside is for early people. Your offer is ₹33 lakh, and here's what the equity is worth under three exit scenarios."

When a candidate's competing offer is above your band

"That's a strong offer and I'm not going to pretend we can match it in cash. Our band tops out at ₹37 lakh. What I can do is move you to the top of the band, or give you a larger grant if you'd rather take more equity. Which matters more to you?"

When a current employee finds out a new hire earns more

"You're right that we brought her in higher. The market for that role moved about 10% since you joined. We're reviewing the whole band now, and I'll come back to you within two weeks with where you sit and what we're changing." (Then do it.)

Answering an investor who asks how you set pay

"We have a written policy: base at the 40th percentile of funded-startup data, equity at the 70th, reviewed every 12 months. Here are the bands and the loaded cost per hire in the plan."

Common mistakes#

1. Benchmarking against the wrong population. IT-services pay understates the market for a strong Indian engineer; big-tech pay overstates it. Use funded-startup data.
2. Negotiating each offer from scratch. This rewards negotiators and creates pay gaps you will have to fix later, often at the worst time.
3. Comparing Indian CTC with a US base salary. Compare loaded cost to loaded cost.
4. Old salary templates in India. A structure with basic under 50% of wages does not comply with the Labour Codes in force since November 2025.
5. Asking for salary history. It is banned in many US states and in EU countries applying the Pay Transparency Directive. Ask what the candidate is looking for instead.
6. Never reviewing the bands. In 2026 some markets moved more than 10% in a year.
7. Ignoring the scientist problem. Engineering benchmarks often underprice PhD scientists with rare skills, and overprice them for roles where a master's degree is enough. See 5.14.

Related playbooks#

• 5.02 ESOP plan design: India vs US
• 5.03 Pool top-ups and who absorbs the dilution
• 5.11 Explaining equity to employees
• 5.12 Trading off cash and equity in offers
• 5.14 Paying PhDs and scientists compared with engineers
• 5.17 Bonuses, commissions and sales compensation plans
• 5.18 Equity for employees in other countries
• 16.03 Budgeting and burn; 16.04 Managing runway

Sources#

• Carta, State of Startup Compensation: H1 2026 (published 28 October 2026)
• Carta, State of Startup Compensation H2 2025 (May 2026) and How AI is changing the compensation game for VC-backed startups (April 2026)
• Pave founding-engineer benchmarks, as cited in Standout, Startup engineer compensation benchmarks 2026
• Ravio, Startup salaries (2026 UK data)
• xto10x, Evolving Compensation Trends (March 2026), as reported by Outlook Business
• Instahyre, Recrew, Hyring (AmbitionBox data), Supersourcing: India salary guides 2026 (aggregator data)
• US Bureau of Labor Statistics, Employer Costs for Employee Compensation (March and June 2026); IRS Publication 15 (2026) for the Social Security wage base
• Payscale, Pay transparency laws: legislative lowdown (May and June 2026)
• Directive (EU) 2023/970; transposition status from Trusaic and Ravio trackers (September 2026)
• Indeed Hiring Lab, salary information in job postings (March 2026)

This playbook is general information, not legal, tax or employment advice. Salary data changes quickly; check current figures and local rules before making offers.