One of seven role studies in the ATS Resume Keyword Study counts 3,518 job postings from 81 companies, term by term. Terms counted in more than one role are compared in the same-term comparison across roles.
Keyword study
Most resume advice about keywords is guesswork. This is a count.
Before you apply
The aggregate says no stack is universal. The posting you are applying to has one. Paste both and see which of its terms your resume never uses.
Check My Resumescalability, mentoring, distributed systems, system design / architecture each appear in more postings than any single programming language. A resume built entirely from a tools list is missing the vocabulary that most postings lead with.
Mentoring appears in 47.4% of postings: more than Kubernetes (30.9%), React (21.1%) or TypeScript (19.8%). It is one of the most commonly asked-for and least commonly written-about things on this list. If you have onboarded, reviewed for, or taught another engineer, it belongs in a bullet.
All 13 languages we counted, their rank among the 45 terms, and whether the gap to the next language is larger than the sampling noise.
| Language | Overall rank | Postings | Share | Next one down |
|---|---|---|---|---|
| Python | 5 | 609 | 44.8% | clearly behind |
| Java | 7 | 482 | 35.4% | clearly behind |
| Go | 8 | 423 | 31.1% | clearly behind |
| TypeScript | 14 | 269 | 19.8% | within noise |
| C++ | 18 | 236 | 17.4% | within noise |
| SQL | 19 | 235 | 17.3% | clearly behind |
| JavaScript | 21 | 179 | 13.2% | within noise |
| Kotlin | 25 | 163 | 12% | within noise |
| Rust | 27 | 135 | 9.9% | within noise |
| Scala | 31 | 117 | 8.6% | within noise |
| Ruby | 32 | 115 | 8.5% | clearly behind |
| C# | 41 | 72 | 5.3% | clearly behind |
| Swift | 45 | 31 | 2.3% | n/a |
No language is close to universal. Python leads at 44.8%, so more than half of these postings never name it. Writing a resume to match “the” software engineering stack is writing to a stack that does not exist.
The top 3 separate cleanly. Python, Java and Go each sit clearly ahead of the next language down. Below that, neighbors are often within the noise, so the order of the middle of this table should not decide what you learn next. List the languages you have shipped in, strongest first.
On the frontend, one framework. React 21.1%, Vue 3.2%, Angular 2.9%. React is named about 7 times as often as Angular.
Share of the 1,360 postings that mention each term, with the posting count in brackets. A posting that says “Python” nine times counts once.
| Term | Postings | Share |
|---|---|---|
| Python | 609 | 44.8% |
| Java | 482 | 35.4% |
| Go | 423 | 31.1% |
| TypeScript | 269 | 19.8% |
| C++ | 236 | 17.4% |
| SQL | 235 | 17.3% |
| JavaScript | 179 | 13.2% |
| Kotlin | 163 | 12% |
| Rust | 135 | 9.9% |
| Scala | 117 | 8.6% |
| Ruby | 115 | 8.5% |
| C# | 72 | 5.3% |
| Swift | 31 | 2.3% |
| Term | Postings | Share |
|---|---|---|
| AWS | 562 | 41.3% |
| Kubernetes | 420 | 30.9% |
| GCP / Google Cloud | 262 | 19.3% |
| Azure | 239 | 17.6% |
| Terraform | 179 | 13.2% |
| Docker | 171 | 12.6% |
| microservices | 97 | 7.1% |
| Linux | 65 | 4.8% |
| Term | Postings | Share |
|---|---|---|
| PostgreSQL | 194 | 14.3% |
| Kafka | 178 | 13.1% |
| MySQL | 153 | 11.3% |
| Spark | 133 | 9.8% |
| MongoDB | 101 | 7.4% |
| Redis | 101 | 7.4% |
| Elasticsearch | 87 | 6.4% |
| Snowflake | 76 | 5.6% |
| Term | Postings | Share |
|---|---|---|
| React | 287 | 21.1% |
| REST / RESTful | 130 | 9.6% |
| GraphQL | 82 | 6% |
| Node.js | 75 | 5.5% |
| Vue | 44 | 3.2% |
| Angular | 40 | 2.9% |
| Term | Postings | Share |
|---|---|---|
| scalability | 719 | 52.9% |
| mentoring | 644 | 47.4% |
| distributed systems | 639 | 47% |
| system design / architecture | 635 | 46.7% |
| cross-functional | 422 | 31% |
| code review | 370 | 27.2% |
| on-call | 281 | 20.7% |
| CI/CD | 265 | 19.5% |
| Agile / Scrum | 127 | 9.3% |
| unit / automated testing | 89 | 6.5% |
Selection. Postings were matched on the title alone, requiring the phrase “software engineer.” Titles like “Senior Software Engineer, Payments” are in; “Engineering Manager” is out.
Language names that are also English words. A bare match on “Go” also catches “go to market” and overstates it by about two points, so it is matched case-sensitively with exclusions. The same applies to React, Spark, Swift and Rust.
No interval column. This was the first study and its rows were recorded without per-term intervals. At 1,360 postings the widest possible is ±2.7 points; the “next one down” column above applies the proper test to each pair.
Limits worth stating. These are 72 mostly US technology companies hiring through one ATS vendor. They are not a random sample of all software engineering jobs, and the mix skews toward well-funded product companies. Terms absent from our 45-term list are not counted, so absence here is not evidence of absence in the market.
adyen · affirm · airtable · anaplan · anthropic · asana · billtrust · blend · braze · brex · carta · checkr · chime · cloudflare · coinbase · coursera · databricks · datadog · discord · doximity · dropbox · duolingo · elastic · figma · fivetran · flexport · gemini · gitlab · gusto · gympass · hellofresh · impact · instabase · instacart · intercom · justworks · klaviyo · lattice · launchdarkly · lyft · mixpanel · mongodb · monzo · newrelic · nuro · okta · oscar · peloton · pendo · pinterest · postman · pushpay · reddit · ridgeline · ripple · robinhood · sagesure · samsara · setpoint · smartsheet · sofi · squarespace · stripe · taskrabbit · tide · tripadvisor · twilio · vercel · verkada · webflow · zocdoc · zscaler
Illustrative pairs, not taken from the postings: the same work, written without and then with the terms postings name.
Before
Built web applications and worked with APIs
After
Architected React + FastAPI platform serving 400k MAU; introduced GraphQL federation layer reducing mobile API payload 60% and cutting Time to Interactive from 4.1s to 1.8s
Before
Worked on backend systems and improved performance
After
Redesigned PostgreSQL query pipeline with covering indexes and materialized views; eliminated N+1 queries and cut P95 latency from 820ms to 95ms for 2M daily active sessions
Before
Set up cloud infrastructure and deployment processes
After
Migrated monolith to 8 Docker microservices on AWS (ECS + RDS + ElastiCache); GitHub Actions CI/CD pipeline ships to production 12× daily with automated rollback on error rate >0.5%
All seven studies use the same collection and counting method, so their figures can be read side by side.
A percentage is the share of the 1,360 postings in this sample that name the term at least once. It is not the share of all employers that ask for it, a cutoff an applicant tracking system applies, or a measure of your chances of an interview.
The 95% interval shows sampling error only. It does not correct for which companies publish on the job boards it draws from or for the one-day snapshot. A larger sample for one role says nothing about which role is more in demand or more important.
The table as it stands: 1,360 postings from 72 companies, 45 terms, method v1.0.
No figure on this page has been corrected. Every withdrawn row and corrected figure across the studies is listed in the corrections log. The analysis script that built this table and the collector it reads from are published as plain text.
Quote any figure on this page. If you do, link back to it so readers can see the sample and its limits rather than the number alone.
General citation
ResumeFitting, “What 1,360 Software Engineer Job Postings Actually Ask For” (method v1.0, data collected 2026-08-29). https://resumefitting.com/software-engineer-job-description-keywords
APA style
ResumeFitting. (2026-08-29). What 1,360 Software Engineer Job Postings Actually Ask For. https://resumefitting.com/software-engineer-job-description-keywords
MLA style
ResumeFitting. “What 1,360 Software Engineer Job Postings Actually Ask For.” ResumeFitting, 2026-08-29, resumefitting.com/software-engineer-job-description-keywords.
Chicago style
ResumeFitting. “What 1,360 Software Engineer Job Postings Actually Ask For.” ResumeFitting, 2026-08-29. https://resumefitting.com/software-engineer-job-description-keywords.
Before you apply
The aggregate says no stack is universal. The posting you are applying to has one. Paste both and see which of its terms your resume never uses.
Check My ResumeOr read the same infrastructure terms, counted in DevOps and SRE postings · Python and Go, counted in security postings
The rest of the research
The study hub holds the method, every role's figures, the limits of each sample, and the full dataset as a CSV.