Back in May 2025, I beat over 2000 applicants to get a 6-figure remote Data Scientist position - with no connections, ivy-league degree or FAANG on my resume. Just cold-applied with 3 years of experience at a smaller private company you've never heard of. And that role too was a remote data analyst position I also got via cold-applying. I want to provide some of the best advice I can give to job-seekers today.
1. Don't Stuff Keywords, Do Understand Requirements
Usually what the employer really wants in a candidate is not going to be written in bold with the exact tech stack and projects they want to put you on. This is too much info to give to competitors and also too much for the candidate to create an ideal but vacuous profile. So they tell you a little bit (e.g. maybe "experience with bots" in an ads data scientist role). But what's really going to get their attention is a bullet point about building a specific bot-detection algorithm. Or they might mention "geospatial experience" and are looking for specific geospatial libraries and GIS technologies. So you need to read between the lines and ask yourself what are they really looking for and what experiences or skills can you highlight that are going to convince them of that? That is the art of building a good resume.
2. Focus on Impact, not Languages/Tools.
I think most people can admit to having been guilty of this at some point, especially as a junior. People will write bullet points like "used python to build ML models". This is about as useful as saying "used lab notebook to do chemistry research". It's awkward because the focus is on a relatively minor implementation detail rather than on the substance. Instead focus on the impact of your work, e.g. "Reduced costs by 20% by designing and implementing A/B tests to optimize offer price". The implementation details will come out in the interview. If you want to stress experience with a particular language or skill, do that in the skills section.
3. Formatting
This is very important - if ATS can't parse it, a human is unlikely to see it. But I also think the complexity of making an "ATS-friendly" resume is overblown. You don't need an ATS "specialist". It's enough to avoid columns, avoid symbols/icons and use a normal resume template. Once you've satisfied that, you can focus on the human experience afterward, wherein lies the real problem. When using AI tools, ChatGPT or Claude tend to destroy your resume template. It will almost certainly still be ATS-parsable, but look sloppy to a human. Here's a huge cheat code. Build your resume in Latex and copy the tex code (or get a resume template in latex online and copy that) and tell your AI tool to use that template. AI is not good at understanding the structure of your template from your pdf, but is very good at understanding and preserving structure from the code that generates the pdf.
4. Apply to more jobs or spend more time on each resume?
This is at the heart of job-seeking philosophy right now. Is it just a numbers game? My honest opinion is that it's a mix of both. I think spending 30+ minutes on any given cold application is doing yourself a disservice. However, spending no time tailoring your resume to the role and just pumping out applications as fast as humanly possible - this is also a waste of time due to the unlikelihood of receiving a response with this approach. I think the sweet spot is about 5-10 minutes on each cold application.
I hope you found this useful. This is actually what led me to build Turquoise Tailoring — a resume tailor that does a lot of what I describe above automatically, and offers the same resume template I used to get my role.
Top comments (0)