BLUEPRINT
$120,230 Data Seat Blueprint: You Have the CS Degree. The Door You're Aiming At Collapsed.
You finished a computer science degree, you want to work with data, and you have sent out applications into what feels like a void. This page is not going to tell you the market is fine. It is going to tell you which number explains what is happening to you, and which four doors are still open.
THE SITUATION
Postings up 47%. Hiring down 73%.
- The number that explains your inbox. Job postings labeled "entry-level software engineer" grew roughly 47% between late 2023 and late 2024, while actual hiring into those roles dropped approximately 73% over the same window. You are applying to requisitions that exist on a board and not in a budget.
- Your major's unemployment is real. The New York Fed puts recent computer science graduates at 7.0% unemployment and computer engineering at 7.8%, against 4.2% across all recent graduates — fifth highest of 73 majors, tied with fine arts and performing arts.
- And the supply side is the quiet half. Early-career computer science graduates went from fewer than 70,000 in 2009 to more than 250,000 in 2024. The market did not only contract. The cohort tripled.
- But read the other column. CS underemployment is 19.1% against 42.5% for all recent graduates. Majors with lower unemployment are frequently achieving it by taking work unrelated to the degree. Your problem is the wait, not the outcome.
WILL AI TAKE THIS JOB?
This is the most exposed seat on this site. Read it anyway.
The ranks. Microsoft Research scored 785 occupations for how much of the work current AI can do. The median occupation sits at #392. Every seat a CS graduate targets for data work sits far above it:
| Data Scientists (15-2051) | #19 |
| Management Analysts (13-1111) | #23 |
| Computer Systems Analysts (15-1211) | #76 |
| Database Architects (15-1243) | #77 |
| Database Administrators (15-1242) | #99 |
| Software Developers (15-1252) | #120 |
| Information Security Analysts (15-1212) | #205 — least exposed of the set |
The physical share is effectively zero across all of them. There is no crawlspace in this work, which is exactly why the scores are what they are.
THREE POINTS OF CONSIDERATION
- It measures what was observed, not what's possible. The score reflects what people were seen asking a general-purpose AI assistant to do, and how much got done. A high rank means that work showed up a lot and got completed.
- It can't see industry tools — but here it doesn't need to. On other seats this is a blind spot. We found it on paralegals, where the score reads low because legal AI lives behind a firm login. Software and data work is the inverse case: the assistants are the tool. For this occupation group the instrument is pointed exactly right, which is why these ranks should be taken seriously rather than discounted.
- It isn't about robots. Irrelevant here. Nothing in this job is physical.
Source: Microsoft Research, "Working with AI" (Tomlinson et al., 2025), CC BY 4.0. Ranks computed by this author from the published score file.
Is This You?
- You hold a bachelor's in computer science or a close relative, you can actually write and ship code, and you want to work with data rather than build product features.
- You have sent a large number of applications to roles titled some variation of "Data Analyst I" or "Junior Data Engineer" and received mostly silence.
- You are willing to take a first job that is not the job you wanted, if you can see the path from it.
All three: this page is written for you. The third one is the hard one and it decides everything below.
NOT YOUR NEXT MOVE IF…
- You are holding out for the title. One researcher's read of the CS unemployment number is that graduates know they can command a high salary and may be holding out longer for the opportunity that pays it. If that is you, understand you are choosing the wait. That is a defensible choice at 22 with no dependents. It is a choice, not a market condition.
- You are counting on more credentials to fix it. About 32.7% of CS majors go on to a graduate degree. A master's is a reasonable move for someone who wants research or a specific specialization. It is an expensive way to wait out a hiring cycle, and it does not address the thing employers are actually screening for, which is evidence you have done the work.
- You want a seat AI won't touch. Look at the table above. If low AI exposure is your priority, this entire occupational family is the wrong neighborhood and the Build-Out Board is the right one.
Where to go instead: if you want to stay technical but move toward the least-exposed end of computing, Information Security Analyst is #205 against data science's #19, pays a $129,180 median, and your degree is a direct qualification.
IF YOU READ NOTHING ELSE — THE 15-MINUTE FIRST MOVE
Stop searching job titles. Search employers.
Open a list of the twenty largest employers in your metro that are not technology companies. Hospital systems, insurers, utilities, banks, manufacturers, state agencies, defense contractors, universities. For each one, go to its careers page and search "developer," "engineer," "analyst," and "data."
Write down two things per employer: whether they posted anything in the last 30 days, and what experience level they asked for.
Why this works. These organizations cannot compete with technology firms for experienced engineers, so they hire people they can train and they keep them for years. They also receive a fraction of the applications a tech company does for the same posting. You are not lowering your standards. You are aiming at the part of the market where your degree is scarce instead of the part where it is the entry requirement.
What It Pays, Reconciled
Four federal numbers, all BLS OEWS May 2025, against an all-occupations median of $50,980:
| Seat | Median | What it tells you |
|---|---|---|
| Software Developers (15-1252) | $135,980 | 1,717,800 jobs growing to 1,892,600 — about 174,700 openings a year. The biggest seat by far. |
| Information Security Analysts (15-1212) | $129,180 | The least AI-exposed seat in this family. |
| Data Scientists (15-2051) | $120,230 10th $67,240 · 90th $199,130 | 275,600 jobs, 35% projected growth — and #19 of 785 on AI exposure. Both of those are true. |
| Computer Occupations, All Other (15-1299) | ~$116,600 10th ~$55,900 | The catch-all where a lot of real data jobs are actually coded. Worth searching. |
Now the number that matters more than any of those. NACE puts projected starting pay for computer science graduates at about $81,535. The gap between that and the $120,230 data scientist median is not a pay cut you are being offered. It is the distance between entry and the middle of a distribution that includes people with ten years in. Treat the medians as the destination and the starting figure as the on-ramp, and do not let a recruiter quote you the median as though it were an entry offer — or let a bootcamp quote it to you as though it were a starting salary.
One honest note on the data scientist spread. The 10th percentile is $67,240 and the 90th is $199,130 — nearly 3x. BLS does not classify those percentile boundaries by career stage, so nobody can tell you from federal data where a first-year sits in that range. Anyone who does is estimating.
Why the Door Closed (Three Facts)
1. The junior rung was the task AI took first. The work that used to go to a first-year — writing the query, drafting the report, cleaning the extract, building the first-pass model — is the work a general-purpose assistant now does acceptably. That is not speculation; it is what the #19 rank measures. The seats above the junior rung still exist, which is why mid and senior analyst roles remain genuinely hard for employers to fill while juniors flood the entry pool.
2. True entry-level data roles were always rare. Most organizations never created many junior-only analyst positions to begin with. In one analysis of 6,877 active data engineering postings in May 2026, only 3% were explicitly entry level — 219 roles nationally. The pipeline everyone is told to enter was narrow before AI and is narrower now.
3. Your competition is not other CS graduates. It is the much larger population of bootcamp and certificate holders applying to the same postings. Your advantage over them is real and under-used: you can read a stack trace, reason about complexity, and ship code that runs in production. Almost none of your competition can. The problem is that a résumé does not transmit that, and the postings you are applying to were never read by a human.
Paths In
| Path | Time to the data seat | Evidence That Converts |
|---|---|---|
| A. Backend or internal software at a non-tech employer Insurer, hospital system, utility, bank, manufacturer, state agency. They hire juniors because they cannot win the experienced candidate. |
12–18 months, then internal transfer. This is the path every recruiter and hiring-trend analysis points to. | A production system you touched, named. "Rewrote the nightly claims extract; cut runtime from 6 hours to 40 minutes" beats any portfolio project, because somebody else can confirm it. |
| B. Cleared work Defense and intelligence contractors sponsor clearances and hire at entry level because the cleared talent pool is small |
Clearance processing is months, and the clearance itself becomes the moat for a decade | An active clearance is the single most portable credential in this field. Once you hold one, you are competing against a few thousand people instead of a few hundred thousand. |
| C. Analytics engineering / BI developer The seat between analyst and data engineer — pipelines, models, dashboards that other people depend on |
Direct entry possible. CS graduates are overqualified for analyst and underqualified for senior engineering; this is the gap. | SQL depth plus one orchestration tool, shown working. Not a certificate — a pipeline somebody else runs. |
| D. Security, then data Information Security Analyst is #205 of 785 and your degree qualifies you directly |
Direct entry; $129,180 median; detection engineering and security data work is a data career with a different title | See the InfoSec blueprint for the credential ladder. |
No licensing in any of these occupations. Certifications matter at the margins and never substitute for a system someone else depends on.
Your First 12 Months
- Month 1 — Change the target, not the volume. Twenty non-tech employers, applications filed. Trigger metric: fewer applications than last month, aimed at employers rather than titles.
- Months 1–3 — Take the adjacent job. Backend, internal tools, integration, reporting, support engineering. Trigger metric: employed, writing code, inside an organization that has a data team.
- Months 3–9 — Automate something nobody asked you to automate. Find the report somebody produces by hand every week and replace it. Trigger metric: one system in production that another person relies on, with a before-and-after number attached.
- Months 6–12 — Get known to the data team before the req opens. Ask the data lead what they are short of. This is the Side Door applied inside a building you already badge into, and internal transfers do not require anyone to post a job.
- The exit trigger. If at twelve months there is no data team, no path to one, and no one who will tell you what it would take — the production experience you now have makes you a different candidate externally than the one who graduated. Apply as someone who has shipped.
THE SCOT FREE TAKE
Postings up 47% and hiring into them down 73% is the whole story, and nobody sending those applications has been told that number. It isn't them. It's a board full of requisitions that were never funded.
The thing I'd want a CS graduate to take from the data is the underemployment column, because it reframes the wait. Nineteen percent against forty-two percent means that when CS graduates land, the job uses the degree. Several majors with better unemployment numbers are getting them by taking work that does not require a degree at all. You are in a slow queue, not a dead end, and those feel identical from inside.
The move is unglamorous and it is the same one this site gives everybody: stop applying to the title and go find the employer. Twenty organizations in your city need somebody who can write code and will never be able to hire a senior engineer. That is where a first job is, and a first job is the only thing standing between you and the seat you actually want.
Do the 15-minute first move today: twenty employers, two notes each. Then get the MoneyZoo LAB Report and instant access to The Side Door Playbook. Subscribe here.
Sources
- Federal Reserve Bank of New York, "The Labor Market for Recent College Graduates," outcomes-by-major table using 2024 American Community Survey data, released February 2026 — computer science 7.0% unemployment and 19.1% underemployment; computer engineering 7.8% and 15.8%; 4.2% unemployment and 42.5% underemployment across all recent graduates (ages 22–27 with a bachelor's or higher); 32.7% of CS majors earning graduate degrees.
- IEEE Spectrum, "The Computer Science Degree Isn't Dead" — entry-level software engineer postings growing roughly 47% between late 2023 and late 2024 while hiring into those roles dropped approximately 73% in the same window.
- Preston Cooper, American Enterprise Institute, as reported by NewsNation — early-career computer science graduates rising from fewer than 70,000 in 2009 to more than 250,000 in 2024; the hypothesis that CS graduates may be holding out longer for high-paying opportunities.
- BLS OEWS, May 2025 — software developers $135,980; information security analysts $129,180; data scientists $120,230 (10th $67,240, 90th $199,130); computer occupations, all other approximately $116,600 (10th approximately $55,900); computer occupations overall $109,470; all occupations $50,980. BLS Employment Projections 2025–35 — software developers 1,717,800 to 1,892,600 with about 174,700 annual openings; data scientists 275,600 jobs with 35% projected growth.
- NACE — projected starting pay for computer science graduates of approximately $81,535.
- Revelio Labs, as reported in 2026 labor market coverage — junior postings in software development and data analysis down as much as 67% since early 2023. DEV Community analysis of 6,877 active data engineering postings, May 2026 — 3% explicitly entry level.
- Microsoft Research, "Working with AI" (Tomlinson et al., 2025), github.com/microsoft/working-with-ai, CC BY 4.0 — ranks of 785 occupations; physical shares and ranks computed by this author from the published score and IWA weight files.
A note on the unemployment figure: both 6.1% and 7.0% are in circulation for recent CS graduates, reflecting 2023 and 2024 American Community Survey vintages respectively. The samples behind major-level estimates are small — one analysis put the 95% confidence interval on the computer engineering figure at roughly 4% to 11% — and discouraged workers who leave the labor force mechanically lower measured unemployment. Separately, computer science and information majors account for only about 30% of workers in computer occupations, so major-level data describes graduates rather than the occupation. This page uses 7.0% and labels the vintage.
Corrections: ScotFree@TheMoneyZoo.com
