Paying PhDs and scientists compared with engineers

1. The core idea#

Deeptech companies hire two kinds of technical people who are priced in different markets. Engineers are priced by the software and hardware industry, where pay is set by competition between employers. Scientists, especially those coming straight from a PhD or postdoc, are priced against academia, where pay is set by fellowship and pay-scale tables. A PhD scientist leaving a postdoc may be delighted with an offer that an engineer of the same age would turn down.

That gap is tempting to exploit, and it is a mistake to rely on it. Scientists find out what engineers earn within months of joining, and the ones you most want to keep will get industry offers once they have a year of startup experience on their CV.

Who this is for: a founder hiring PhD scientists, postdocs and research engineers alongside software, hardware and process engineers, typically in India, the US or the UK.

The principles this playbook works from:

• Pay for the level of the job, not for the degree. A PhD is evidence of skill, not a pay grade. Put scientists and engineers on one levelling ladder with equivalent levels (section 3).
• Benchmark scientists against industry R&D, not against academia. The academic number is the candidate's floor, not your market rate.
• Use the academic gap honestly at the first hire. It is reasonable to pay a first scientist below industry rates if you explain it, close the gap with equity, and commit to reviewing pay after the next round.
• Watch the scientists with engineering skills. Computational scientists, ML researchers and anyone who can write production code are priced as engineers, often at a premium.

Related: 5.13 (salary benchmarks), 5.12 (cash vs equity), 6.03 (the deeptech hiring mix), 6.07 (judging research talent), 6.12 (hiring from big pharma and national labs), 2.10 (negotiating with a university).

2. What the data says (November 2026)#

The academic-to-industry gap is large in every field#

The US Survey of Earned Doctorates (NSF, published February 2026, 2024 graduates) gives expected median starting salaries for new PhDs by sector:

Two things matter for a founder. First, a new PhD taking a postdoc earns roughly half of what the same person would earn in industry. Second, the industry premium varies a lot by field: computer science PhDs are priced like senior engineers, while physical and biological science PhDs start much lower.

US industry pay for scientists by level#

Aggregator data for biotech in Boston and Cambridge (2026) gives typical base salary ranges by background:

In computational biology, a PhD adds about $20k–$30k to starting pay, but a master's degree plus 3–4 years' experience often matches a new PhD's starting salary (CompBioJobs, 2026). ML and AI roles in life sciences pay more at every level than

equivalent non-AI science roles, and ML researchers at US frontier startups command $230k–$280k base (EPM Scientific, August 2026).

The pattern: experience in industry is priced more highly than the degree itself, and computational skill is priced more highly than wet-lab skill.

India#

In India the gap between academia and industry is narrower than in the US for wet-lab and materials scientists, because Indian industry R&D pay is itself low. A postdoc on ₹80,000 a month (₹9.6 lakh a year, often with a research grant on top) may only be 20–40% below an entry industry scientist. The big gap opens for computational and AI scientists, who are priced against software and big-tech research labs.

United Kingdom#

• UKRI's minimum PhD stipend for 2026/27 is £21,805, tax-free, from 1 October 2026.
• Postdoctoral research associates are paid on university pay spines, typically about £40,000–£46,000 at entry (Grade 7), with London weighting on top.
• A mid-level startup software engineer earns about £70,800 at seed (Ravio, 2026; see 5.13). A UK postdoc moving to a startup often sees a 25–50% raise at an offer well below the engineering median.

What this means#

• The first scientist hire is the cheapest point of the gap. That is when an honest below-market offer, plus equity, is most defensible.
• The gap closes within 2–3 years. After that, your scientists' outside market is industry, not academia. Budget for it.
• Computational and ML scientists are not in the academic market at all. Price them using engineering data (5.13).

3. How to do it: one ladder, separate market bands#

The approach most growing deeptech companies settle on is: one set of levels for everyone, separate salary bands for each job family, and equity set by level, not by family. This keeps the internal logic fair while still paying what each market requires.

Step 1. Build one levelling ladder with parallel tracks

The level descriptions must be about scope and judgment, not credentials. That is what lets you place a strong master's-level scientist at L3 and a fresh PhD at L2 without arguments.

Step 2. Decide how to count academic experience#

There is no standard rule, so pick one and apply it to everyone. A common approach: count the PhD as the entry ticket to L2, and count postdoc years at about half the weight of industry years when deciding on L3. Override it when the postdoc work is directly relevant (the same assay, the same materials system, the same instrument), because then it is worth as much as industry experience.

Step 3. Price each family against its own market#

Use the 5.13 method for each job family separately:

• Wet-lab, materials and process scientists: industry R&D data for your sector and city, not academic pay scales.
• Computational scientists, ML researchers, bioinformaticians: engineering data. They are hired by the same employers as software engineers.
• Hardware and process engineers: engineering data for your sector.

Step 4. Set equity by level, not by family#

Give the same equity range to a L3 scientist and a L3 engineer. This is the main tool for fairness when the cash bands differ, and it signals that you value the science as much as the engineering. (See 5.02–5.03 for pool sizing.)

Step 5. Decide on a maximum gap and write it down#

If the science band at a level sits well below the engineering band, write down the reason and the rule, for example: "At the same level, science base pay may be up to 30% below engineering base pay where the market requires it. Equity is identical. We review this at every fundraise." A written rule is much easier to defend than an unexplained gap someone discovers.

Step 6. Plan the catch-up#

A first scientist hired from a postdoc will be priced by industry in two to three years. Budget for a step-up in pay at 12–24 months, or at the next round, and say so in the offer.

Worked example#

A synthetic biology startup in Bengaluru has raised a ₹ 25 crore seed round. Policy from 5.13: base at the 40th percentile for each family, equity by level. It is hiring three people at the same time.

What the founder does about the gap between A and C. They are at the same level, and A earns a third less. The founder tells A directly: the band is set by the market for biotech R&D in India; A's equity grant is identical to C's; and A's pay will be reviewed against the industry market after 12 months, when A will have industry experience. The founder writes the 30% maximum-gap rule into the compensation policy.

What changes in the US. A new PhD Scientist I at a seed startup in Boston would benchmark at about $115k–$145k against a postdoc salary of about $60k–$65k (NSF). At the 40th percentile, an offer of about $125k nearly doubles the candidate's pay, but they will check it against biotech offers, not against their postdoc. A computational biologist would benchmark against ML engineering pay, often $160k–$210k.

4. Country notes: visas, fellowships and joint appointments#

For scientists, pay decisions interact with immigration and academic rules more than they do for engineers. Check these before you set an offer.

United States: pay now affects your H-1B odds#

• The H-1B lottery is weighted by wage level. Under the DHS final rule published on 29 December 2025 and effective 27 February 2026, each registration gets one to four entries depending on the Department of Labor wage level the offered salary meets (Level I = 1 entry, Level IV = 4). Early modelling suggests selection odds of roughly 15% at Level I against about 61% at Level IV.
• The practical consequence: an offer at the bottom of the market for a new PhD may fall into wage Level I and sharply reduce the chance of an H-1B. For an international scientist you need, check the prevailing wage level for the occupation and county before you fix the salary. Paying at Level II or III can be the difference between getting the hire and not.
• The $100,000 fee on certain new H-1B petitions for workers outside the US (presidential proclamation, September 2025) has been subject to litigation. Changeof-status petitions for people already in the US, such as F-1 students, have generally been exempt. Check its status with immigration counsel at the time of hiring.
• Cap-exempt routes: universities and nonprofit research organisations are exempt from the cap. A scientist who keeps a part-time appointment at a cap-exempt university can sometimes work concurrently for a startup without entering the lottery. This needs a genuine university role and careful structuring.
• O-1A: often the better route for experienced scientists with publications, patents or awards. There is no lottery and no wage-level rule, but the evidence bar is high. See 1.06 (founder visas).

United Kingdom: the PhD salary discount#

The Skilled Worker route lets you pay less where the person has a relevant PhD:

The discounts cannot be combined. Pay at the discounted level only if it is also fair against your bands; the person will need to meet the full going rate later for settlement. The Global Talent visa has no salary floor and is often better for strong researchers, because it is not tied to your company. The Migration Advisory Committee began a review of Global Talent in May 2026, so check for changes.

India: fellowships, government scientists and joint appointments

• Fellowship holders usually cannot take a salaried job alongside the fellowship. Most national fellowships (for example ANRF's National Post-Doctoral Fellowship, INSPIRE Faculty) are full-time and tied to the host institution. A fellow who joins you normally gives up the fellowship. Your offer should beat the fellowship plus its research grant, not just the stipend.
• Faculty can often stay part-time. Many IITs, IISc and other institutions allow faculty to consult, hold equity or take leave to work with startups, within their own rules. A professor can be a part-time chief scientist rather than an employee. See 1.03 (leave and dual roles), 2.12 (professor as co-founder, advisor or CSO) and 3.10 (part-time and academic co-founders).
• Government scientists (CSIR, DRDO, ISRO) have pension and job-security benefits that matter more than headline pay. Some labs allow deputation or entrepreneurial leave. Expect to pay a premium to compensate for giving up security.
• Grants can fund scientist salaries. BIRAC BIG and other early-stage grants often allow salary costs for research staff. This can let you pay a scientist closer to industry rates without spending equity money (see 13.01 (map of non-dilutive money) and 13.03 (BIRAC: BIG, SPARSH, PACE)).

Everywhere: publication and IP rights#

Scientists often value the right to publish and present as part of their package. Writing a clear publication policy (what can be published, after what review delay, with what patent filing first) costs you nothing in cash and can matter more to a candidate than a 10% raise. Make sure the IP assignment in their employment agreement covers it (4.09).

5. Scripts, mistakes and sources#

Scripts

Making an offer to a postdoc

"We pay scientists against the industry R&D market, not against academic pay, and this offer is in our band for that market. It's a big step up from your fellowship, but I want to be straight: a pharma company might offer you more cash. What we offer instead is the same equity as an engineer at your level, the chance to own a programme, and a pay review after 12 months, when your market becomes industry rather than academia."

When a scientist asks why an engineer at the same level earns more

"That's a fair question. We use one set of levels and identical equity for everyone at a level, but the cash bands follow two different markets, and the software market currently pays more. Our policy caps the gap at 30%, and we review it at every fundraise. Here's where you sit in your band and what it would take to move up."

Pricing a computational scientist

"For this role we're using engineering benchmarks rather than science benchmarks, because the companies competing for you are hiring ML engineers."

Explaining a publication policy

"You can publish your work. We review drafts for patentable material and aim to file within 60 days so publication isn't held up for long."

Common mistakes#

1. Benchmarking scientists against academic pay. The postdoc stipend is the candidate's floor, not the market. You will lose them within two years.
2. Paying for the degree, not the job. Putting every PhD above every master's-level scientist ignores the industry experience that the market actually rewards.
3. Different equity for scientists and engineers at the same level. This makes an existing cash gap feel like a judgment about whose work matters.
4. Pricing computational scientists as scientists. They are in the engineering market and will leave for it.
5. Ignoring visa wage rules. A low US offer can drop an H-1B registration into wage Level I and cut its selection odds; a UK offer below the discounted threshold cannot be sponsored at all.
6. Forgetting the fellowship grant. Fellows often lose a research grant as well as a stipend when they join you.
7. No plan for the catch-up. If you hire below market, say when and how you will review pay, and do it.

Related playbooks#

• 5.12 Trading off cash and equity in offers
• 5.13 Salary benchmarks for startups in India and the US
• 5.02–5.03 ESOP plan design and pool top-ups
• 6.03 Scientists, engineers or operators: the deeptech hiring mix
• 6.07 Judging research talent: publications vs ability to ship
• 6.12 Hiring from big tech, big pharma and national labs
• 1.03 Leave, sabbatical or dual role; 1.06 Founder visas
• 2.12 Professor as co-founder, advisor or CSO; 3.10 Part-time and academic co-founders
• 4.09 Employment contracts, NDAs and invention assignment
• 13.03 BIRAC: BIG, SPARSH, PACE

Sources#

• NSF NCSES, Expected median salaries for doctorate recipients are highest in the industry sector (NSF 26-312, February 2026; Survey of Earned Doctorates 2024) JobRise, Biotech jobs Boston Cambridge 2026: salary (aggregator data)
• CompBioJobs, bioinformatics and AI drug discovery salary data (2026)
• EPM Scientific, Machine learning salaries in life sciences: 2026 compensation guide (August 2026)
• Dheya, Research scientist salary in India 2026 ; AcademicJobs India fellowship and biotech salary articles (2026)
• UKRI minimum doctoral stipend 2026/27 (via CASRAI, 2026); university pay spines (2026)
• GOV.UK, Skilled Worker visa: when you can be paid less ; Danielle Cohen Immigration and Reiss Edwards summaries of the July 2025 thresholds
• DHS final rule on weighted H-1B selection (published 29 December 2025, effective 27 February 2026); BAL, Grossman Young and Davis Wright Tremaine summaries (2026); RN Law Group modelling of selection odds
• Gherson, MAC review of Global Talent and Innovator Founder routes (May 2026)

This playbook is general information, not legal, tax, immigration or employment advice. Visa rules in particular change often; check with counsel before making an offer that depends on sponsorship.