Home/Blog/Careers & Students
Careers & Students · 10 min read

How to land a software or data job as a new graduate in 2026

The short answer

Treat the search as a pipeline you can measure. Choose one or two role lanes you can evidence, decode each posting, send a one-column resume tailored in minutes, show two interview-ready projects, get referrals through real conversations, practise coding patterns and SQL out loud, and review your funnel every week. Then compare offers on first-year value.

Want the tools, not just the method? The The Tech and Data Job Search Kit kit turns this guide into ready-to-use templates, spreadsheets and printables, with an instant download.

Key takeaways
  • Pick one or two lanes; generic applications get generic results.
  • Plain, one-column resumes parse well; evidence bullets win the human read.
  • Moderately weak ties produced the most job moves in a 20-million-person experiment.
  • Being watched lowers interview performance: practise thinking aloud.
  • Review your funnel weekly and change one thing.
  • Compare offers on first-year value, not base pay.

What are the steps to land a software or data job?

The steps below work for new graduates, bootcamp graduates and career changers. Each one has a measurable output, so you can tell which part of your search needs work. The order matters: lanes come before resumes, because the lane decides which keywords, projects and people matter, and practice comes before the interview invitations arrive, because the gap between an invitation and the interview is often only a few days. A meta-analysis of 47 job search programmes found that the ones that worked combined skills with motivation (goals, proactivity and support from other people), so the weekly review is part of the method, not an extra.

  • Read the market and choose one or two lanes you can evidence.
  • Build a target list of about 30 employers and decode postings before you apply.
  • Rebuild your resume in a one-column, plain format and tailor it per posting.
  • Make two or three projects interview-ready with READMEs and tests.
  • Network in small, specific conversations; ask for referrals only when a role fits.
  • Track every application and review your funnel weekly.
  • Practise coding patterns, SQL and short design questions out loud, with a timer.
  • Prepare eight STAR stories, then compare offers on first-year value.

Which tech roles are growing, and what do they pay?

Start with official data rather than social media. The U.S. Bureau of Labor Statistics projects software developers to grow 10% from 2025 to 2035, with about 106,100 openings a year across developers, QA analysts and testers, and a median pay of $135,980 for developers in May 2025. Data scientists are projected to grow 35% with a median of $120,230.

Two cautions. A median covers all experience levels, so look at lower percentiles: in Raleigh-Cary, North Carolina, software developers earned $103,480 at the 25th percentile and $80,080 at the 10th. And the market for new graduates is tighter than the averages: the New York Fed put recent-graduate unemployment at 5.2% in the third quarter of 2026, with 42.5% working in jobs that do not need a degree.

How do I choose a lane?

A lane is a role family you can evidence, such as junior backend engineer or data analyst. Score each candidate lane on four questions: what proof you can show today, the gap a hiring manager would see, how many openings exist near you or remote, and whether you want the work.

Choose one or two lanes that share at least half their skills, so your projects, keywords and practice serve both. Pairing a stretch lane with an adjacent one (software engineer with data analyst or QA) is common and sensible. Review the choice after six weeks of tracking: if one lane gets replies and the other does not, that is information. Career changers should look for lanes where their past field is an advantage, such as a teacher moving into education data or an accountant moving into finance analytics.

Do applicant tracking systems reject most resumes?

An applicant tracking system is the software employers use to receive applications, store resumes and move candidates through stages; it parses your file into fields that recruiters search and filter. The popular claim that 75% of resumes are rejected by software before a human sees them has no sound study behind it that we could find. What is documented is that employers configure filters, and some filters screen out capable people: Harvard Business School's Hidden Workers research found rules such as excluding career gaps removed applicants before anyone read them.

So write for two readers. Use one column, standard headings (Skills, Experience, Projects, Education), no tables, images or text boxes, and plain-text links. Then put evidence first: an action verb, the task, the scale and the result. Mirror a posting's exact terms only where they are true.

Do referrals and networking really help?

The strongest evidence comes from randomised experiments on LinkedIn's 'People You May Know' feature covering more than 20 million people over five years. Moderately weak ties, people you know a little, produced the most job moves, and weak ties mattered more in digital industries.

Make networking small and specific: two or three 20-minute conversations a week about someone's work, a thank-you the same day, and a referral ask only when the role fits. Track your response rate by source; in the worked example below, applications made through people got far more replies. Alumni of your university or bootcamp, people who spoke at a local meet-up and people who wrote publicly about their team are the easiest first contacts. Prepare five questions, listen more than you talk, and end by asking who else you should speak to.

Worked example: what does a measured search look like?

This example follows Andre Brooks, a fictional 2026 computer science graduate in Raleigh, through a 12-week search after an unfocused summer. The numbers are illustrative but internally consistent, and they show how to read a funnel. Read the table from top to bottom. He sent fewer applications than in the summer, but almost a third of them went through people, and the response rate rose from under 3% to about 17%. Referred applications were also his best-matched ones and the samples are small, so the lesson is not a rate to expect but a habit: count replies by source every week and move hours towards what works.

MeasureSummer (unfocused)12 weeks (method)
Applications14084
Through referrals or warm introductions025
Responses (screen or later)4 (2.9%)14 (16.7%)
Response rate, referral or warmn/a9 of 25 (36%)
Response rate, cold2.9%5 of 59 (8.5%)
Technical interviews, final rounds, offers0, 0, 08, 4, 2
Measure where you are

Reading helps; measuring tells you what to work on. These AI-graded assessments on AssessAll pair with this topic:

How should I prepare for coding interviews?

Practise by pattern rather than by random problem: hash maps, two pointers, sliding windows, stacks, binary search, prefix sums, intervals, heaps, breadth-first and depth-first search, simple dynamic programming and parsing. Two patterns a week, a 10-minute attempt before any hint, and a re-solve two days later builds skill you can show.

Practise out loud with someone watching. In a randomised study at North Carolina State University, students solving a whiteboard problem while observed performed less than half as well as those working alone. Use a six-step loop every time: clarify, examples, plan aloud with the cost, code while narrating, test, and discuss trade-offs. If you want an outside measure of interview readiness, AssessAll's Job Interview Readiness assessment is one option.

Online assessments often come first. Read the invitation for the time limit, languages and rules on outside resources, try the platform's sample question, and plan your time per problem. Hidden test cases check edge cases, so test empty inputs, duplicates and large values before you submit. Afterwards, log the patterns that came up and practise them the following week.

What about SQL, system design and behavioural interviews?

Data roles usually test SQL: joins, grouping, filtering groups, missing values, window functions and dates. Say what one result row means before you write a query, and check row counts after every join, because joins multiply rows. For case questions, such as 'sign-ups are up 20% but revenue is flat', clarify the metric, check the data for duplicates and tracking changes, break it down by plan, channel and cohort, offer two or three hypotheses with the data that would test them, and recommend one next step.

Junior system design questions reward sensible simplicity: ask for numbers first, sketch the simplest design that fits, then say what you would change at ten times the load. A useful outline takes about 35 minutes: clarify users and numbers, list the main API calls, define the tables and the main query, sketch client, server and database, grow it for more load, and name trade-offs and risks such as failure, abuse and privacy. Behavioural interviews are predictable because structured interviews ask about past behaviour; the U.S. Office of Personnel Management's guide describes scoring answers against examples. Prepare eight STAR stories of about two minutes, each with a number in the result. Spend about half of each answer on your own actions, say 'I' rather than 'we', and rehearse aloud with a timer. Most behavioural questions fall into about a dozen themes, such as conflict, failure, ownership, a deadline, learning fast and feedback, so eight strong stories can cover nearly all of them.

Common mistakes in a tech job search

  • Applying to every role family with one resume.
  • Treating every listed requirement as a must-have.
  • A designed two-column resume that parses badly.
  • Asking strangers for referrals in the first message.
  • Measuring effort (applications sent) instead of results (replies by source).
  • Grinding problems silently without re-solving.
  • Using AI during an assessment that forbids it.
  • Accepting an offer on the phone before seeing written terms.

How do I compare job offers?

Compare first-year value, not base pay: add base, signing bonus, the expected bonus (not guaranteed), employer retirement match and other employer cash, then subtract your share of health premiums and commuting. Then weight what matters to you, such as growth, money, fit, flexibility and stability, and score each offer. In the worked example, a US$92,000 engineering offer with a signing bonus and a 4% match came to US$103,300 in the first year, against US$86,200 for a US$80,000 remote analyst offer with a larger learning budget. Weighted for growth, money, fit, flexibility and stability, the engineering role scored 4.2 out of 5 and the analyst role 3.4.

Negotiate once, politely, with a reason and a specific number, and be ready to accept a no. Good reasons are a written competing offer, a published pay range or a skill the team said it needed. Ask about the whole package (start date, signing bonus, learning budget, remote days) as well as base pay, and get the final terms in writing before you resign from anything. If you need help keeping the weekly habits going, Jobulary's guide to if-then plans is a practical read. The Tech and Data Job Search Kit includes an Offer Comparison Calculator and a decision memo for this step.

Bodhih Pro Kit

Run the whole search with one set of files

The Tech and Data Job Search Kit gives you the role map, tracker, resume templates, practice pack with tests, interview scorecards and offer calculator in one place.

Sources

  1. U.S. Bureau of Labor Statistics, Occupational Outlook Handbook: Software Developers, Quality Assurance Analysts, and Testers
  2. U.S. Bureau of Labor Statistics, Occupational Outlook Handbook: Data Scientists
  3. O*NET OnLine (U.S. Department of Labor, Employment and Training Administration), BLS OEWS data: Local wages: Software Developers 15-1252.00, North Carolina
  4. Federal Reserve Bank of New York: The Labor Market for Recent College Graduates
  5. U.S. Office of Personnel Management: Structured Interviews: A Practical Guide
  6. ACM ESEC/FSE 2020 (NSF Public Access Repository): Behroozi, Shirolkar, Barik and Parnin: Does stress impact technical interview performance?
  7. Science (PubMed record): Rajkumar, Saint-Jacques, Bojinov, Brynjolfsson and Aral: A causal test of the strength of weak ties, 377(6612), 1304-1310
  8. Psychological Bulletin (Monash University research portal): Liu, Huang and Wang: Effectiveness of job search interventions, a meta-analytic review, 140(4), 1009-1041
  9. Harvard Business School, Working Knowledge: How to tap the talent automated HR platforms miss (Hidden Workers study, Fuller and colleagues)
  10. National Bureau of Economic Research (published in Management Science, 2025): Wiles, Munyikwa and Horton: Algorithmic writing assistance on jobseekers' resumes increases hires (w30886)
  11. U.S. Federal Trade Commission: Job Scams
  12. Journal of Applied Psychology (via Singapore Management University repository): Sackett, Zhang, Berry and Lievens: Revisiting meta-analytic estimates of validity in personnel selection, 107(11), 2040-2068
Common questions

Questions people ask next

How many applications should I send each week?

Fewer, better-matched applications usually beat volume. Many people aim for 8-12 tailored applications a week with two or three through people, then adjust after four weeks of tracking response rates by source.

Is a one-page resume required?

For US new graduates, one page is the norm. In the UK and Ireland the document is a CV and two pages is common with experience. Either way, keep one column and plain formatting.

Do I need a portfolio for a data analyst role?

It helps. One analysis that starts from a real question and messy data, shows the cleaning and ends with a recommendation is often more convincing than several tutorials.

Can I use ChatGPT or Claude to write my resume?

AI tools can tighten wording, and a field experiment found writing help on resumes led to more hires. Never let a tool add skills or numbers you cannot explain, and do not paste private data into it.

Is it cheating to use AI in an online coding assessment?

If the instructions say no outside resources, yes. Use AI only where the employer allows it in writing, and disclose it when asked.

How long does a tech job search take?

It varies widely with the market, your lane and your evidence. Plan a structured 12-week effort with weekly reviews rather than expecting a fixed timeline.

What are signs of a job scam?

Being asked to pay for a job or training, to deposit a cheque and send money back, an offer with no real interview, or requests for bank details before a written offer. Report scams to the platform and, in the US, the FTC.