The levels look standardised and are not. SDE1, SDE2 and SDE3 came from a handful of large product companies and then spread across the Indian market with the definitions loosened at every step. Two candidates with the same title on their CV can differ by several years of real scope.
This is what the levels actually mean, what each costs in 2026, and where the definitions break.
What each level actually means
Years of experience are the worst proxy, and the one most job descriptions use. Scope and independence are what separate the levels.
SDE1 — delivers. Ships well-specified work with review. Writes tests, works inside existing patterns, asks before diverging. Typically 0–2 years. What you are assessing is whether they learn quickly and whether their code needs rewriting.
SDE2 — owns. Takes a feature or service end to end, including its design, and is trusted to make reasonable calls without checking. Breaks down ambiguous work themselves. Typically 2–5 years, and the level where most engineers spend the longest stretch of their career. This is the level most teams actually need and the one they most often mis-hire, because a strong SDE1 interviews similarly and costs less.
SDE3 / Senior — decides. Owns systems other teams depend on, sets technical direction within a team, and makes decisions that are expensive to reverse. Mentors by default. Typically 5–8 years. The test is not whether they can build it, but whether their judgment about what to build has been right before.
Staff / Principal — directs. Technical strategy across teams, usually without direct reports. The work is as much organisational as technical — the decisions cross team boundaries and the hard part is agreement, not implementation.
Salary bands, 2026
Total compensation, including equity where it applies. Bengaluru, Hyderabad, Pune and NCR are broadly comparable at the top end; tier-2 cities typically run 20–30% below.
| Level | Product company | Funded startup | GCC | IT services |
|---|---|---|---|---|
| SDE1 | ₹16–28 LPA | ₹14–26 LPA | ₹18–30 LPA | ₹6–12 LPA |
| SDE2 | ₹28–50 LPA | ₹25–45 LPA | ₹30–55 LPA | ₹12–22 LPA |
| SDE3 | ₹50–85 LPA | ₹45–80 LPA | ₹55–95 LPA | ₹22–40 LPA |
| Staff | ₹85 LPA–1.6 Cr | ₹75 LPA–1.4 Cr | ₹90 LPA–1.5 Cr | ₹40–65 LPA |
Three things these bands do not show.
The tier gap is larger than the level gap. An SDE2 at a product company can out-earn an SDE3 at a services company. If you are hiring from services into product, the offer will look enormous to the candidate and ordinary to your existing team — check your internal parity before you send it.
Equity is not comparable. A startup offering ₹30 LPA plus equity and a GCC offering ₹45 LPA cash are not straightforwardly rankable, and candidates increasingly discount equity heavily. If you are competing on equity, be specific about the strike price, the current preference stack and the last round's valuation. Vagueness reads as evasion.
AI and ML roles carry a premium of roughly 20–40% over these bands at the same level, and more for people with production experience on large models. That premium has been rising for two years.
What each level costs to hire
Compensation is the visible cost. These are the rest.
| Level | Time to fill | Recruiting fee (15–25%) | At 8.33% |
|---|---|---|---|
| SDE1 | 3–6 weeks | ₹2.4–7 L | ₹1.3–2.3 L |
| SDE2 | 4–8 weeks | ₹4.2–12.5 L | ₹2.3–4.2 L |
| SDE3 | 6–12 weeks | ₹7.5–21 L | ₹4.2–7 L |
| Staff | 10–16 weeks | ₹12.7–40 L | ₹7–13 L |
Then add what the process costs you internally: roughly 42 hours of senior-engineer time per role across résumé triage, screens, interviews and debriefs. At SDE3 salaries, that is a meaningful number on its own — and it is spent before anybody is hired.
And add the notice period. 30 days is common at SDE1 and SDE2; 60 to 90 days is standard at SDE3 and above. An offer signed today at SDE3 commonly means someone starting in three months.
Where the level definitions break
Title inflation at small companies. A four-person startup may give an SDE3 title to a second-year engineer, honestly and reasonably — relative to that team, they are senior. It does not transfer.
Services-to-product. Years in a services company often map to fewer years of product scope, because the work is specified elsewhere. This is not a judgment about capability; it is a difference in what the job involved. Assess the scope directly.
The 5-year cliff. The largest capability spread in the market sits between four and seven years' experience. Some people at five years are genuinely operating at SDE3; others are doing SDE1 work with more confidence. Title and years both fail here, which is why a structured rubric matters more at this band than anywhere else.
How to assess level properly
Stop mapping titles. Ask what they owned.
- "What is the largest thing you've owned end to end? Who else touched it?"
- "Tell me about a design decision you made alone that turned out wrong." SDE2s and above have one. SDE1s usually describe a decision someone else made.
- "What in your current system would you rebuild, and what would break if you did?"
- "What did you disagree with your tech lead about, and what happened?"
The pattern you are listening for is independence under ambiguity. That is what separates the levels, and it is not visible on a CV.
A note on hiring at these levels
SDE1 and SDE2 roles usually fill from applications — those candidates are genuinely in the market, and a good job post plus a structured process will work.
SDE3 and above usually will not. The pool is smaller, and strong senior engineers are rarely applying anywhere: across our own placements, 72% of the people hired were not looking when we found them. At that level, if you are only reading applications you are choosing from the minority of the market that happens to be available right now.
That is the point where outbound search starts to be worth its fee — and where it does not, we will tell you. We work software engineering searches across India, the US and Europe on a success fee of 8.33% of first-year salary, paid on the day the person joins. Send us a role if one is stuck.
Bands are compiled from offers made and benchmarked across roles closed to date, plus publicly reported market data. They are indicative — validate against your own market before making an offer.