A resume should explain the business problem, the work performed, and the result, rather than list responsibilities and technology keywords.
2
Interview preparation starts with researching the company, the interviewers, and the role, then preparing specific examples with the STAR structure.
3
Candidates improve their odds by practising aloud, explaining their individual contribution, asking engaged questions, and making clear why they want the job.
Summary
Anthony Kelly explains how data scientists and machine learning engineers can get more opportunities from the same job search. He recommends tailoring resumes to the role, starting with recent and relevant experience, and describing projects from the original business problem through implementation and measurable outcome. Technology lists should show depth and context instead of claiming many tools without examples. He also covers education, publications, cover letters, LinkedIn, and direct outreach for targeted applications. For interviews, Kelly recommends researching the company and interviewers, preparing STAR answers around the job description, and explaining personal contributions rather than hiding behind the team. Candidates should practise by recording themselves, prepare questions, show genuine interest in the company, and describe mistakes together with what they changed afterwards. His advice is practical and grounded in recruitment experience, with a clear focus on giving candidates several options rather than relying on one offer.
A resume needs to connect work to business results
Kelly says the experience section is the most important part of a CV. It should begin with the most recent and relevant role, include the company name and a short explanation of what the company does, then describe responsibilities and accomplishments. A project should explain the problem, what the candidate built, and what changed for the business. His examples include a sales forecasting model with a 25 percent improvement and an expected business impact of 50 to 100 million per year. The point is to make the candidate's contribution understandable to someone who may not know the company or its domain.
Kelly warns against listing many programming languages or tools without showing how they were used. A data scientist with two years of experience who lists six languages can raise questions about their actual depth. He prefers candidates to describe skills through projects, including how they integrated systems, deployed code, worked with data pipelines, or connected tools to the business. This can also help people moving from software engineering into data science or machine learning. Software experience becomes useful when the resume explains how it relates to data, infrastructure, deployment, efficiency, and the needs of a data team.
Recent graduates should use education to explain their problem-solving
For recent graduates, education belongs near the top because it is often the strongest part of the CV. Kelly recommends listing relevant secondary degree courses and explaining the topics in more detail. Graduate and intern resumes often describe what the person built without explaining the challenge they set out to solve. He suggests writing a CV in a way that resembles an interview answer, so the reader can understand the problem, the candidate's decisions, and the result. Publications can be included, but they should support the application and may deserve more emphasis when they directly match the company's work.
Targeted applications need a tailored cover letter and direct outreach
For a small set of target companies or a highly specialised role, Kelly recommends writing an individual cover letter for each position. The letter should show that the candidate understands the company, identify challenges suggested by its work, and explain how the candidate could help. He also recommends finding the hiring manager's contact details and writing directly instead of relying only on a general inbox. This approach is more suitable for a dream company or a narrow market than for sending the same level of personalisation to every possible employer.
A plain, readable CV is better than a decorative template
Kelly advises against downloading a CV directly from LinkedIn because it can look like a lack of effort. He prefers a simple layout with a readable font, the candidate's name and introduction, personal information, education, and career history. He does not recommend side panels, charts, or visual skill bars. LinkedIn still matters because recruiters may review it before contacting someone, including their experience, posts, talks, and other activity. The LinkedIn profile and the CV can use the same descriptions, as long as both explain the work clearly.
Kelly says the best-prepared candidate often gets further than the candidate with the strongest raw profile. Before an interview, candidates should read the company's website, look for recent articles and investment news, review the interviewer's LinkedIn profile, and understand the role. This helps them discuss the company's work in specific terms and gives them material for relevant questions. Research also signals interest. Kelly argues that interviewers want to know whether the candidate likes the company as well as whether the company likes the candidate.
STAR answers should explain the candidate's own contribution
Kelly recommends preparing answers with the situation, task, action, and result structure. Candidates should select important topics from the job description and prepare a concrete example for each one. A strong answer explains the challenge, the team context, the candidate's responsibility, the technologies involved, and the outcome. Saying only that the team built something leaves the interviewer unsure about the individual's ability. Kelly also warns against using internal language, since interviewers may not understand an internal tool or process without an explanation of its purpose and value.
Practice and engaged questions improve interview performance
Kelly recommends recording practice interviews and listening back to find unclear explanations, repeated words, and nervous habits. He compares this with athletes practising regularly before a game. Candidates should prepare ordinary questions about the company and role, then listen for details during the interview and ask follow-up questions about them. One question he suggests is how the company would judge success after six months in the role. After the interview, a short message explaining what interested the candidate can keep the conversation open and make their interest clear.
Mistakes are useful when the answer shows what changed
When asked about challenges or mistakes, candidates should not claim they have never failed. Kelly recommends describing a real mistake, the consequences, and what the candidate did to improve. He gives an example of someone trying test-driven development after watching a webinar, finding that the first attempt failed, and then working with a specialist on the next project. The useful part of the story is the willingness to learn and the specific change in behaviour. A failure answer should show reflection and follow-through rather than defensiveness.
"A good recruiter will always ask why aren't you interviewing that CV and why are you interviewing that CV."Anthony Kelly10:32
Who should watch
You are applying for data science, machine learning engineering, data engineering, analytics, or related roles and your resume mainly lists duties and tools.
You are moving from software engineering into data science or machine learning and need to connect your existing work to data-team problems.
You have interviews coming up and want a repeatable way to research companies, prepare examples, practise answers, and ask better questions.
