JavaScript, Python, and SQL still dominate US developer hiring. They don't compete for the same seats.
Want a language that matches real postings, not a trophy list?
Start with jobs you can apply to this week. Then learn the language those ads already name. LinkedIn, Indeed, and Dice make that check fast if you search by stack instead of scrolling another hot-skills roundup.
Job posts and developer surveys measure different things
Thing is, a popularity poll is not a requisition.
The 2025 Stack Overflow survey, as summarized across several recaps, still has JavaScript as the language most developers touch at work. SQL and Python sit right behind it. TypeScript keeps showing up as the typed layer on JavaScript for product teams.
Oxylabs analyzed nearly 1 million U.S. job postings and got another order. Python was named most often, in about 46% of the tech ads in that set. SQL appeared almost as often, usually next to another language. Java landed near 21%. JavaScript was around 19%.
Use the survey when you want to know what coworkers already speak. Use the posting scrape when you want to know what an ATS might catch. You need both.
What to target instead of a fake number-one
Evidence is mixed, so a single national rank would be theater. Compare languages by the work they unlock.
| Language | What the data usually shows | Stronger fit | | JavaScript and TypeScript | Highest everyday use in developer surveys | Frontend, full-stack web, many remote product roles | | Python | Most-named language in one large US posting study | Data, ML, automation, plenty of backend work | | SQL | Near-constant co-mention with other languages | Analytics, data engineering, anything that queries production data | | Java | Heavy in enterprise and Android hiring | Banks, internal systems, Android apps | | Go or Rust | Lower volume, often senior or systems-shaped ads | Cloud services, infrastructure, performance-sensitive code | | Swift or Kotlin | Native mobile clusters | iOS (Swift) or Android (Kotlin) |
PHP still feeds a lot of web employment because WordPress runs a huge share of sites. It just doesn't trend well.
C and C++ work is real and narrow. Games, embedded devices, trading engines, operating systems. If those products don't interest you, skip them.
Match the stack to the posting pile you can reach
Open Indeed or LinkedIn first. Search the language. Read ten ads.
Most web product ads want React or Next.js with TypeScript, not "coding" in the abstract. Browser work still runs through JavaScript. Serious product roles now assume TypeScript too. That's the lane for a lot of remote frontend and full-stack searches.
Python shows up when the work is data, scripting, or machine learning. Recaps of US ads have said AI-related wording in postings rose more than 130% over several years, and that a large share of data and analytics ads now mention AI. Python is the language those ads usually list. It's not a magic interview. Pair it with SQL or you stall on the first take-home.
Java is the language you learn because the company already runs on it. That's boring. It also explains why Java keeps a fat slice of enterprise postings even when it isn't the survey darling. Android work still leans on it, with Kotlin in the mix.
SQL rarely hires you by itself. You'll still fail data screens without it.
Go and Rust attract people who like systems work. Volume is smaller. Titles skew senior. Rust also polls as highly admired, which is a compliment, not a headcount report.
To be honest, I would not spend a year on Rust because a survey called it beloved if your local boards are full of Java and SQL.
You start with Python because every list screams AI, then you grind notebooks, then you open Indeed and a pile of the jobs want Java, Spring, and a clearance you don't have, and three months later the portfolio still doesn't match the ads in your city. That's the expensive version of following a rank.
Hiring is uneven, and junior volume is the tight part
BLS still lists software developers in the Occupational Outlook Handbook. Check official wages and outlook there before you trust a screenshot of somebody's offer. Data scientist figures live on a separate BLS page if that's the posting pile you're chasing.
Editorial roundups of posting indexes described a market only a little above pre-pandemic baselines, with developer ads in some analyses still well below that older peak. A lot of any rebound sat in senior titles. AI mentions climbed. Entry-level generalist ads did not climb with them.
Specialization still beats collecting syntax. A Python resume with no SQL and no domain project looks like a course, not a hire.
Run the job boards like a filter, not a slot machine
Dice, LinkedIn, and Indeed all work. They work better when each saved search maps to one stack.
Don't spray 80 Easy Apply clicks at "software engineer." Recruiters screen for the language and the tools around it.
Do this:
- Choose one target stack for the next 30 days. TypeScript plus a web framework, Python plus SQL, or Java plus the library your local ads repeat.
- Create two or three LinkedIn saved searches. One for your metro area. One for remote. One with a tight title. Refresh filters, because old result sets stick around.
- Mirror those keywords on Indeed. Write down the extra tools that keep appearing: AWS, Docker, Tableau, Android, Spark.
- Apply where you already match most listed skills. A short daily list beats a hundred identikit applications.
- After roughly 40 applications with no conversations, change the stack or the proof on the resume. Tweaking only the summary line rarely fixes a mismatch.
Company career pages still matter. Many employers prefer that path over the in-network button. Dice is useful when you already know the language and want contract or niche volume, not as your only board.
Proof employers can open in under a minute
Your GitHub and resume have to survive an ATS keyword pass and a 30-second human skim.
Keep the language names identical to the posting. If they say TypeScript, don't hide it under "frontend." If they say PostgreSQL, don't only write "databases."
Build one project that looks like the job:
- Web: a TypeScript app with a real UI, auth, and an API
- Data: Python plus SQL against a messy dataset, with a chart you'd show a manager
Enterprise Java is the same idea in a different costume. A small service with tests and a README that sounds like work will beat a language tour.
Mobile stays its own market. Swift for iPhone work. Kotlin for Android. Cross-platform names appear, but native ads still cluster.
Freelance gigs follow the same JavaScript and Python spread you see on full-time posts. The contracts are shorter. The keyword game is the same.
Stop at one language until interviews start
Skip the third language. You don't get extra credit for a graveyard of unfinished tutorials.
Spend four weeks on the stack that already has ads you could apply to this week. Ship something. Put the same nouns in the README that employers put in the requisition.
Turns out the useful test is simple. Search your chosen language on LinkedIn tonight. If you can't find roles you'd actually submit, pick the language with the bigger pile and start there.
Count the ads. Count them again after you add the framework those ads name.