Role ranking video: hire flexibility
A recent video that ranks software roles argues companies often hire new grads into adjacent engineering buckets—backend, platform, or data—and then pivot them toward ML work, rather than hiring straight into senior ML research slots. The piece emphasises transferable SWE fundamentals like APIs, testing, and infra as common entry paths. (youtube.com)
Machine learning jobs often start with plain software work: companies hire into backend, platform, or data teams, then move people closer to models later. (youtube.com) The video, “i ranked every SWE role,” says compensation data came from Levels.fyi and demand projections came from the United States Bureau of Labor Statistics’ 2024-2034 outlook, published in August 2025. Its argument is that entry paths are wider than the job title “machine learning engineer” suggests. (youtube.com) That lines up with how major artificial intelligence labs describe the work. OpenAI says its research engineers need “solid engineering skills” for massive distributed machine learning systems, and Anthropic tells candidates that engineers there “do lots of research” while researchers “do lots of engineering.” (openai.com, anthropic.com) The underlying job is less about inventing a model from scratch than keeping a model system running, like plumbing and power for a building. OpenAI’s machine learning engineer posting lists distributed compute, data orchestration, storage, streaming, and machine learning infrastructure as core work. (openai.com) Google uses similar language. Google DeepMind says software engineers scope, build, maintain, and upgrade systems for machine learning, and its research engineers act as a bridge between theory and implementation. (deepmind.google) That helps explain why backend habits travel well. Application programming interfaces, tests, deployment, monitoring, and debugging are the tools companies use to turn a model demo into a product that survives real traffic. (openai.com, anthropic.com) The hiring market also shows more than one front door. Simplify’s live new-grad tracker listed 364 entry-level roles overall, including 51 in data science, artificial intelligence, and machine learning, while its software engineering section was far larger at 252 roles. (github.com) Some companies still post direct machine learning jobs for graduates. Indeed’s listings page showed hundreds of “new graduate machine learning engineer” openings this week, including roles at Nvidia and ByteDance, but many of those postings ask for coursework, advanced degrees, or specialized machine learning experience. (indeed.com) The senior end is narrower. Anthropic’s current machine learning systems engineer role asks for 4 or more years of software engineering experience, and Google lists early-career artificial intelligence and machine learning software roles separately from staff and research positions. (job-boards.greenhouse.io, google.com) So the practical ranking in the video is less about one “best” title than about mobility. Get hired where teams need reliable software builders, and the path toward machine learning work usually runs through the code, tests, and infrastructure first. (youtube.com, google.com)
