Freelance Data Work Under Presumptive Tax vs a Salary
In this article (5 sections)
Related programme: the Generative & Agentic AI course (3 months).
Under the presumptive scheme in section 58(2) of the Income-tax Act, 2025, an eligible freelancer in a specified profession can declare 50% of gross receipts as profit. The receipts limit is ₹50 lakh, or ₹75 lakh where cash receipts are 5% or less. On hypothetical receipts of ₹24,00,000, deemed profit is ₹12,00,000. That is fully covered by the new-regime rebate in tax year 2026-27, so the tax is nil. A ₹24,00,000 salary owes ₹2,92,500. The tax gap is real, but it is not the whole comparison: a freelancer carries costs, idle months, GST and no employer benefits.
Who the section 58 scheme covers
Section 58(2), Table Sl. 3 of the Income-tax Act, 2025_Bill.pdf) replaced the old section 44ADA. It applies where all of the following hold:
- you are a resident individual or a firm, not an LLP;
- your profession is one specified in section 62(4), which lists information technology and technical consultancy;
- gross receipts are within ₹50,00,000, or ₹75,00,000 if cash receipts are no more than 5% of the total.
Deemed profit is 50% of gross receipts, or a higher figure if you claim one. Under section 58(4), no separate deduction for expenses or depreciation is allowed on top.
Whether a particular data engagement is "information technology" or "technical consultancy" is a question of fact about your work. If you are unsure, ask a chartered accountant before relying on the scheme.
The same ₹24,00,000, two ways
Both columns use the new-regime rates for tax year 2026-27 (1 April 2026 to 31 March 2027). The freelance figures are hypothetical.
| Line | Salary | Freelance, presumptive |
|---|---|---|
| Gross | ₹24,00,000 | ₹24,00,000 receipts |
| Deduction or deemed profit | standard deduction ₹75,000 | profit at 50%: ₹12,00,000 |
| Total income | ₹23,25,000 | ₹12,00,000 |
| Tax with cess | ₹2,92,500 | ₹0 |
| Employee PF (₹3,000 × 12) | −₹36,000 | – |
| Real costs you bear (laptop, software, internet, desk) | – | −₹1,80,000 |
| Cash after tax and costs | ₹20,71,500 | ₹22,20,000 |
| Employer PF credited to your EPF | ₹36,000 | – |
| Paid leave, group insurance | usually included | you buy your own |
The freelance advantage shrinks from ₹2,92,500 of tax to ₹1,48,500 of cash once costs and PF are counted. It also assumes twelve billed months. Two idle months cost ₹4,00,000 of receipts, and the freelancer comes out behind.
The ₹24,00,000 edge
Deemed profit reaches ₹12,00,000 exactly at ₹24,00,000 of receipts. Above that, marginal relief applies. Tax cannot exceed the income above ₹12,00,000, plus cess, until the normal slab takes over at about ₹12,70,588 of income.
| Receipts | Deemed profit | Tax with cess |
|---|---|---|
| ₹24,00,000 | ₹12,00,000 | ₹0 |
| ₹25,00,000 | ₹12,50,000 | ₹52,000 |
| ₹26,00,000 | ₹13,00,000 | ₹78,000 |
| ₹28,00,000 | ₹14,00,000 | ₹93,600 |
The step from ₹24,00,000 to ₹25,00,000 of receipts costs ₹52,000 of tax, so you keep only ₹48,000 of the extra ₹1,00,000. The rebate is not a flat allowance; plan projects near the line with this in mind.
# Synthetic receipts and costs. New regime, tax year 2026-27.
def slab_tax(ti):
t, start = 0.0, 0
for end, r in [(400000, 0), (800000, .05), (1200000, .10), (1600000, .15),
(2000000, .20), (2400000, .25), (10**12, .30)]:
t += max(0, min(ti, end) - start) * r
start = end
return 0.0 if ti <= 1200000 else round(min(t, ti - 1200000) * 1.04, 2)
def salaried(gross, employee_pf=36000):
return gross - employee_pf - slab_tax(gross - 75000)
def freelance(receipts, costs, billed_months=12):
earned = receipts * billed_months / 12
assert earned <= 5000000, "check the ₹50 lakh / ₹75 lakh receipts limits"
return earned - costs - slab_tax(0.5 * earned)
print("salary:", salaried(2400000))
print("freelance, 12 months billed:", freelance(2400000, 180000))
print("freelance, 10 months billed:", freelance(2400000, 180000, 10))
for r in (2400000, 2500000, 2600000, 2800000):
print(r, slab_tax(0.5 * r))Things presumptive tax does not change
- Regime choice works differently. With business or professional income, section 202(4) requires the old-regime option by the return due date. Once chosen it continues, and it can be withdrawn only once. A salaried person can choose each year with the return.
- Clients may deduct TDS under section 393 at 10% on professional fees, even though your final tax is nil. That is up to ₹2,40,000 of cash you wait to recover. See client TDS on freelance invoices.
- GST is a separate test. Section 22 of the CGST Act requires registration once aggregate turnover in a financial year exceeds ₹20 lakh (₹10 lakh in special category States). ₹24,00,000 of receipts would cross it. GST registration for freelance data work tracks that threshold month by month.
Records worth keeping from the first invoice
The scheme rests on three facts you may have to show later. Your receipts stayed within the limit. Your cash share stayed within 5% if you rely on the ₹75 lakh limit. And your work falls within a specified profession. Keep a simple ledger with the date, client, invoice number, amount, mode of receipt and TDS deducted for every invoice. Keep the contract or statement of work, because it describes the service. A ledger kept monthly takes minutes; rebuilding one at filing time takes days.
Many people now take on applied GenAI projects as freelancers alongside, or instead of, a salaried role. Our Generative & Agentic AI add-on is a 3-month programme, live online with optional in-person labs at our Noida centre.
Frequently asked questions
What replaced section 44ADA?
In the Income-tax Act, 2025, presumptive taxation for specified professions is in section 58(2), Table Sl. 3. It applies to a resident individual or firm (not an LLP) in a profession listed in section 62(4), which includes information technology and technical consultancy.
Can I claim my laptop and software costs as well?
Not on top of the presumptive figure. Section 58(4) says no further deduction or allowance is allowed once income is computed this way. The 50% is meant to cover your costs.
Pankit Kumar has 10 years in Data Science & AI, building and shipping production systems in regulated pharma and clinical environments. He is a freelance trainer at Boston Institute of Analytics, AnalytixLabs and Scaler, and has taught this material to thousands of working professionals.
This article is part of our Generative & Agentic AI programme — 3 months. Add practical GenAI, retrieval and agent-building skills to your existing toolkit.
Explore Generative & Agentic AI