Resume Building for Data Roles: A Complete Guide

Resume building for data roles is a matching exercise. A screening system or a recruiter compares your page with a job posting in seconds, and a hiring manager then looks for evidence that you can do the work. Hiring processes built to find "perfect" candidates efficiently end up systematically excluding many qualified people, as research from Harvard Business School on "hidden workers" found (HBS: Hidden Workers). A resume that matches the language of the job and proves results gets past both readers. To ground the advice, the code below reads the US Department of Labor's O*NET data (O*NET 31.0 database) for three data occupations.

from collections import Counter

import pandas as pd

# O*NET 31.0: technologies per occupation; "In Demand" = classified as in demand for that occupation
tech = pd.read_csv("onet/Software Skills.txt", sep="\t")
tasks = pd.read_csv("onet/Task Statements.txt", sep="\t")
roles = {"15-2051.01": "BI analyst", "15-2051.00": "data scientist", "15-1243.01": "data warehousing specialist"}

demand = {}
print("R1 technologies O*NET marks as in demand, by role")
for code, name in roles.items():
    t = tech[tech["O*NET-SOC Code"].eq(code)]
    demand[name] = set(t.loc[t["In Demand"].eq("Y"), "Workplace Example"])
    print(f"   {name} ({len(demand[name])} of {len(t)} listed): " + ", ".join(sorted(demand[name])))

common = set.intersection(*demand.values())
print(f"\nR2 in demand for all three roles: " + ", ".join(sorted(common)))

print("\nR3 verbs that open each role's official task statements (most common first)")
for code, name in roles.items():
    verbs = Counter(task.split()[0].rstrip(",") for task in tasks.loc[tasks["O*NET-SOC Code"].eq(code), "Task"])
    print(f"   {name}: " + ", ".join(f"{v} ({n})" for v, n in verbs.most_common(8)))

R1 technologies O*NET marks as in demand, by role
   BI analyst (14 of 203 listed): Amazon Web Services AWS software, Microsoft Azure software, Microsoft Excel, Microsoft Office software, Microsoft Power BI, Microsoft PowerPoint, Python, R, SAP software, SAS, Salesforce software, Snowflake, Structured query language SQL, Tableau
   data scientist (21 of 87 listed): Amazon Web Services AWS software, Apache Hadoop, Apache Spark, C++, Git, Microsoft Azure software, Microsoft Excel, Microsoft Power BI, Microsoft PowerPoint, NumPy, PyTorch, Python, R, SAS, Scikit-learn, Snowflake, Structured query language SQL, Tableau, TensorFlow, The MathWorks MATLAB, pandas
   data warehousing specialist (10 of 159 listed): Microsoft Access, Microsoft Excel, Microsoft Office software, Microsoft Outlook, Microsoft Power BI, Microsoft PowerPoint, Operational Data Store ODS software, Python, Structured query language SQL, Tableau

R2 in demand for all three roles: Microsoft Excel, Microsoft Power BI, Microsoft PowerPoint, Python, Structured query language SQL, Tableau

R3 verbs that open each role's official task statements (most common first)
   BI analyst: Maintain (2), Identify (2), Create (2), Analyze (2), Generate (1), Manage (1), Provide (1), Document (1)
   data scientist: Identify (3), Apply (2), Analyze (1), Create (1), Test (1), Deliver (1), Recommend (1), Compare (1)
   data warehousing specialist: Develop (3), Create (3), Design (2), Verify (1), Map (1), Provide (1), Write (1), Perform (1)

Which tools to put on the page

O*NET classifies some technologies as "in demand" for each occupation, separately from its "hot technologies" that appear frequently across job postings (O*NET: Software Skills). Six were in demand for all three data roles: SQL, Python, Excel, Power BI, Tableau and PowerPoint. Beyond those, BI analysts' lists included cloud platforms (AWS, Azure), Snowflake, SAP, Salesforce and SAS. Data scientists' lists added pandas, NumPy, scikit-learn, PyTorch, TensorFlow, Spark, Hadoop and Git. For data warehousing specialists, the closest O*NET occupation to a data engineer, the in-demand list was short and focused on SQL, Python and reporting tools.

Two lessons follow. First, SQL and Python appear everywhere, so list them only if you can prove them, and prove them in your experience bullets, not only in a skills list. Second, PowerPoint and Excel are in demand for all three roles because data work ends in communication. Do not leave them off because they feel basic.

Match the posting, honestly. If a job asks for Power BI and you have used Tableau, say Tableau, and mention any Power BI practice you really have.

Which verbs describe the work

Official task statements show how the work itself is described. Data warehousing tasks most often began with develop, create and design; data scientist tasks with identify, apply, analyse and recommend; BI analyst tasks with maintain, identify, create and analyse. Use verbs like these when they are true, because they describe outcomes rather than duties. "Designed", "built", "automated", "identified" and "recommended" beat "responsible for" and "worked on".

A structure that works

1. Header: name, city, email, phone, LinkedIn and GitHub or portfolio links.
2. Summary (two lines): your target role, your strongest evidence and your domain. "Data analyst with two years in retail operations; built the weekly sales dashboard used by 40 store managers."
3. Skills: grouped and short (languages, databases, BI, cloud), no ratings or bars.
4. Experience: reverse chronological, three to five bullets per role, results first.
5. Projects: essential for freshers and career switchers; two or three with links.
6. Education and certifications: brief, unless you are a recent graduate.

Keep it to one page for under about eight years of experience, and two at most beyond that.

The bullet formula

Action + what you did + tool + measurable result.

• Weak: "Responsible for creating reports in Power BI."
• Strong: "Automated the monthly sales report in Power BI and SQL, cutting preparation from two days to two hours."
• Weak: "Worked on a churn model."
• Strong: "Built a churn model in Python (scikit-learn) that flagged 300 at-risk accounts a month; retention offers to them were adopted by the sales team."

If you do not have an exact number, use scale (rows, users, reports, teams) or a before-and-after. Never invent numbers; interviewers will ask about them.

Formatting for screening software

• Use a simple single-column layout with standard headings: Experience, Projects, Skills, Education.
• Save as PDF unless a site asks for Word.
• Avoid tables, text boxes, icons and images for important content; some systems read them badly.
• Write tool names as postings do: "SQL", "Power BI", "Amazon Web Services".

Tailoring for each application

Keep one master resume with every project and result you could use, then make a short tailored copy for each application. Reorder bullets so the most relevant evidence comes first, adjust the summary to name the role and domain, and swap in the project closest to the employer's work. Tailoring takes about fifteen minutes and usually does more than sending many identical copies.

Checklist before you send

• Does the top third show your target role and best evidence?
• Does every bullet show a result or scale?
• Are the posting's main tools present, and true?
• Are all links working, and do your projects have clear READMEs?
• Has someone else read it for typos?

Pair the resume with a short cover letter (how to write a cover letter that gets read) and a matching LinkedIn profile. For entry-level candidates, see your first data job with no experience.

Portfolio projects that strengthen a data resume are part of the Data Analytics course.

Sources

• Software Skills file, O*NET 31.0 data dictionary (O*NET Resource Center) — O*NET Resource Center
• O*NET 31.0 Database (O*NET Resource Center) — O*NET Resource Center
• Hidden Workers: Untapped Talent (Harvard Business School, Managing the Future of Work) — Harvard Business School