The numbers in that job listing breakdown tell a quiet story, and it's worth reading carefully. For the roles jacobnar is targeting, the data shows a clear tilt: Cohere's MTS Post-Training position lists 48% Masters and 22% PhD, while the Research Engineer role sits at 52% Masters and 26% PhD. Even the Associate Applied Scientist role at Garner Health, the most accessible of the three, still shows 56% Masters. The question isn't whether a BS gets hired. It's whether the degree is a filter or a signal, and the honest answer is that it's both. This tension between formal credentials and demonstrated capability is exactly what we explored in How AI is empowering developers to rebuild familiar software from scratch, where the tools themselves are lowering the barrier to entry for complex work.
A Master's degree is not a hard requirement, but it functions as a soft one in ways that feel almost algorithmic. The hiring managers who publish these breakdowns aren't secretly favoring degrees; they're using them as a proxy for research maturity. A BS holder with a conference paper or a strong project portfolio is competing against a Masters candidate who spent two years doing thesis work. That's not a level playing field, but it's also not an insurmountable one. The 22% to 30% of BS/other hires in those listings are real people, and they got there because their work spoke louder than their credentials. The question jacobnar should ask isn't "will a BS ever get hired?" but "what does my evidence of research capability look like outside of a degree?" Projects matter, but they matter most when they demonstrate depth, not just breadth. A workshop paper, a reproducible open-source contribution, or a deployed model that solved a genuinely hard problem carries weight precisely because it's harder to fake than a transcript.
What's more telling is the way this pattern mirrors a broader shift in how technical roles are being evaluated. The tools we use are becoming more accessible, as Dots Promises a Smarter Workflow, but Data Scientists Should Look Closer points out, yet the gatekeeping around advanced roles remains stubbornly traditional. That contradiction suggests the degree requirement isn't about the material itself; it's about signaling persistence and exposure to a certain kind of thinking. For someone already working full-time as a Data Scientist, the practical path forward is to treat every project as a miniature thesis. Document your process, publish your findings, and make your work visible in places hiring managers actually look. The Jev Projects bring clarity to AI decision-making approach of breaking down complex decisions into traceable steps is exactly the kind of rigor that stands in for formal training.
The takeaway here is direct: if you have a BS and want those roles, stop asking permission and start building a public record of your research process. One strong, well-documented project that shows how you think through model design and failure analysis is worth more than a degree you'd have to pause your career to earn. The 30% and 22% slots exist; they're just not handed out by default. The question is whether your portfolio makes you impossible to ignore.