How few entry-level roles there are
On TixelJobs, 1,933 of 50,447 live listings are entry-level, under 4% of the total. Mid-level roles number 22,822 and senior roles 11,278. Demand runs the other way: 32% of apply clicks since February 2026 went to entry-level roles. That is roughly eight times the share of listings, and it is the shape of the problem. There are not many doors, and everyone is at the same ones.
It is worth knowing this before you start, because it changes the strategy. Applying to every entry-level listing you see is a low-yield use of time. Making yourself the obvious candidate for a few, and finding roles that are entry-level in substance but not in title, works better.
What entry-level listings ask for
Read twenty entry-level ML listings and the same requirements repeat: a degree in computer science, statistics or a related field; Python; one deep learning framework; SQL; and evidence that you have built something end to end. Many also ask for one to two years of experience, which is contradictory and common. Internships, research assistantships and substantial project work are what companies mean by it.
- A degree or a bootcamp plus a portfolio. A master's is asked for more often in research-adjacent roles than in engineering ones.
- One project you can talk about for 30 minutes: the data, the baseline, what you tried, what improved and why.
- Software fundamentals: git, tests, a working knowledge of Linux and a cloud provider.
- Communication. Entry-level engineers spend a lot of time explaining what they did to people who did not do it.
Build a portfolio that survives a screen
The portfolio is doing the job that experience would do, so it has to look like work rather than coursework. Three finished projects beat ten notebooks. Each should have a README that states the problem, the result with a number, and how to run it. A deployed demo, even a small one, shows you can get past training.
Pick problems close to the roles you want. If you are aiming at NLP, fine-tune and evaluate a small model on a real dataset and write up what you found. If you are aiming at computer vision, do the same with a detection or segmentation task. Reproducing a paper and documenting where your numbers differ from the published ones is an underrated project because it is exactly what junior researchers and engineers are asked to do.
Internships and the roles beside the door
Internships are the traditional entry point, and there are 174 internship listings on TixelJobs at the moment. They are competitive and seasonal, and they are covered in the AI internships guide, but the short version is: apply early in the cycle, apply to many, and treat a research assistant post at your university as equivalent.
The routes that get more people in are the adjacent ones, because the numbers are better:
- Data engineer: 6,549 listings on TixelJobs. Data engineers build the pipelines ML teams depend on, sit next to them, and move across regularly.
- AI annotation and data quality: 1,083 listings. Labeling, evaluation and quality work is often genuinely entry-level, and it teaches what good data looks like from the inside.
- Analyst and data scientist roles: 1,414 data scientist listings. Analytics roles are a common first step, and a year of SQL and experimentation transfers well.
- Software engineer on an ML product. Companies hire far more general engineers than ML engineers, and an engineer already on the team is first in line when an ML role opens.
A practical application plan
With this ratio of applicants to openings, the process matters as much as the credentials. A plan that fits in an hour a day:
- Pick two titles and two adjacent titles. Open the hubs for each and check them daily; entry-level roles fill fast.
- Apply within a day of a role appearing. With this many applicants per opening, being early is one of the few advantages you control.
- Tailor the first third of your resume to each role. The project descriptions and the skills line should echo the listing.
- Track outcomes. If twenty applications produce no calls, the resume is the problem, not the market. Change it before sending twenty more.
- Keep building while you apply. A project shipped this month is a better talking point than a course finished last year.
Where to look next
The entry-level AI jobs hub on TixelJobs collects the 1,933 entry-level listings, and the AI internships hub holds the 174 internships. For the adjacent routes, search data engineer and annotation in the full search, and try the ML engineer jobs hub with the level filter set to entry. The remote AI jobs page is worth watching too, since remote roles widen the pool of companies you can apply to. Browsing is free. Membership is $9 a year or $4.90 every three months and unlocks the apply link and the full description on every job. Cancel any time from your billing page.