Analyzing and interpreting data; Framing and synthesis. [28]

Machine Learning Scientist

What this job does:

Research and build state-of-the-art ML methods (often including LLMs), prototype and evaluate models, and partner with engineering and domain experts to deploy robust ML systems or data products.

This job is similar to these lab/research tasks:

Research: implement new methods, compare baselines, run ablations; Benchmarking: curate datasets, define metrics, cross-validate; Engineering: refactor into reusable modules, write tests, optimize runtime; Collaboration: co-design requirements with collaborators, present results; Other: write manuscripts, mentor trainees.

Typical starting salary range

$153k to $250k

Salary estimate represented as 25th to 75th percentile. Gathered from recent job postings, Glassdoor.com, and Salary.com

Daily task breakdown

Model/algorithm research & prototyping 35%
Data curation, experimentation & benchmarking 20%
Engineering/productionization (robust code, integrations) 20%
Stakeholder collaboration & communication 15%
Oother (publishing, mentoring, roadmap) 10%.

Skills you'll use in this job:

Strong ML fundamentals (modeling, optimization, evaluation); Python and ML frameworks (PyTorch/TensorFlow); experimentation and benchmarking; software engineering for reproducibility; ability to translate goals into ML problems and communicate trade-offs.

Talk about these PhD skills in your resume and interview:

Treat your PhD as ML R&D: literature-to-implementation, rigorous baselines/ablations, error analysis, reproducible experiments, and writing/presenting results. Emphasize translating ambiguous questions into formal objectives and shipping usable tools, not just papers.

Alternative job titles to search:

ML Research Scientist; Applied Research Scientist (ML); AI Scientist; Research Scientist (Machine Learning); LLM Scientist; Applied Machine Learning Scientist.

*Note:

-Job titles are inconsistent between companies. Try various keywords.
-PhDs tend to undershoot seniority level. Target higher than you think.