Careers in AI: roles, skills, and paths

Work around AI reaches well beyond research labs. U.S. government occupational guides and employer postings describe researchers, software and machine learning engineers, data specialists, the engineers and technicians who design chips and run data centers, and people in product, design, policy, law, safety, and evaluation. One AI job title often maps to several official occupations, so the duties matter more than the name. Preparation differs by role: research jobs often expect a graduate degree, many engineering jobs a bachelor's degree, and electricians usually train through apprenticeships, which the Department of Labor also recognizes for some data roles. Course materials from universities, government, and companies, and the catalog's Jobs page, can help you explore. This page does not discuss pay or hiring outlook.

Job titles and official occupations

Two Department of Labor resources describe occupations in plain terms. The Bureau of Labor Statistics' Occupational Outlook Handbook (OOH) explains what workers do and how they prepare. O*NET OnLine (external site: onetonline.org), from the Department's Employment and Training Administration, lists tasks, skills, and sample job titles. AI titles rarely match one entry: O*NET's search for "machine learning engineer" (external site: onetonline.org) lists Data Scientists, Computer Systems Engineers/Architects, and Computer and Information Research Scientists first among its closest matches, and Data Engineer appears as a sample title under Database Architects (external site: onetonline.org). Read a posting's duties, then find the occupation that fits them.

Research, software, and data roles

• Research scientists. The OOH (external site: bls.gov) says computer and information research scientists design innovative uses for new and existing technology; their duties include designing experiments, often with data science and machine learning techniques, and publishing and presenting findings. An Ai2 robotics research scientist posting (external site: job-boards.greenhouse.io) describes research on foundation models for robotics, work in simulation and with real robots, mentoring interns, and writing papers.
• Software and machine learning engineers. Software developers create applications and the systems that run devices or control networks, and quality assurance analysts and testers design and run tests (OOH (external site: bls.gov)). Anthropic's careers page (external site: anthropic.com) says its engineers do much research and its researchers do much engineering.
• Data scientists and data engineers. Data scientists collect, clean, and analyze data and create, validate, and test models (OOH (external site: bls.gov)). O*NET lists Data Engineer under Database Architects, who design databases and data warehouses; an Anthropic data engineer posting (external site: job-boards.greenhouse.io) describes building pipelines that turn raw logs into standard datasets and asks for SQL and Python expertise.

Chips, infrastructure, and data centers

An accelerator is a chip built to speed up AI computation, and a data center houses servers with the power and cooling they need.

• Chip and hardware engineers. Computer hardware engineers design and test processors, circuit boards, and memory devices (OOH (external site: bls.gov)); electrical and electronics engineers do similar work on electrical equipment and systems (OOH (external site: bls.gov)). A Cerebras chip design posting (external site: jobs.ashbyhq.com) centers on RTL (register-transfer level) design, which describes a chip's logic in code, with verification, physical design, software, and systems teams.
• Infrastructure and reliability engineers. Network and computer systems administrators install, configure, and maintain networks, operating systems, and servers (OOH (external site: bls.gov)). A Cerebras site reliability engineer posting (external site: jobs.ashbyhq.com) lists releases, capacity changes, cluster upgrades, monitoring, and post-mortems, using Kubernetes, Python or Go, and Prometheus.
• Data-center technicians. A CoreWeave posting inviting interest in technician roles (external site: coreweave.com) describes hardware and network diagnostics, physical repairs, installing equipment, scripting server updates, on-call shifts, and fully on-site work.
• Electricians. Electricians install, maintain, and repair power, lighting, communications, and control systems (OOH (external site: bls.gov)). A Crusoe journeyman electrician posting (external site: jobs.ashbyhq.com) on its manufacturing team covers installation, troubleshooting, and blueprint reading, and requires a Colorado journeyman license.

Product, policy, legal, safety, and evaluation

• Product and design. Ai2's careers page (external site: allenai.org) listed product manager and product designer roles beside research roles. An Ai2 product designer posting (external site: job-boards.greenhouse.io) describes design, engineering, product, and partnerships working beside an ML research team, and visiting users to watch them work.
• Policy and safety. An Anthropic safeguards policy posting (external site: job-boards.greenhouse.io) says the role needs no coding but does require engaging with technical material such as classifiers, turning safety work into policy, and working with government stakeholders.
• Legal. Lawyers typically need a law degree and a state license, usually by passing a bar exam (OOH (external site: bls.gov)).
• Evaluation. Evaluation means measuring how a model behaves, often with a benchmark. An Anthropic evaluations posting (external site: job-boards.greenhouse.io) describes deciding which measurements reveal something real, monitoring evaluations during training runs, investigating regressions, and building dashboards, and lists statistics and experimental design as useful background.

Education and other routes in

According to the OOH pages above, research scientists typically need a master's degree or higher, though some federal jobs accept a bachelor's; software developers, data scientists, and hardware and electrical engineers typically need a bachelor's; and some employers of systems administrators ask for a postsecondary certificate or associate's degree instead.

Registered Apprenticeship combines a paid job with a mentor, classroom instruction, and a nationally recognized credential; you apply through the employer or program sponsor (apprenticeship.gov (external site: apprenticeship.gov)). Most electricians train in a 4- or 5-year apprenticeship, and most states require a license. O*NET lists Data Scientist and Machine Learning Data Curator as Department of Labor-approved apprenticeship titles under Data Scientists (external site: onetonline.org), and the Department runs an AI in Registered Apprenticeship (external site: apprenticeship.gov) portal. For two-year colleges, the National Science Foundation's Advanced Technological Education (external site: nsf.gov) program supports the education of technicians for high-technology fields.

Employers describe other signals too. Anthropic says it values what candidates can do over where they learned it, and suggests putting independent research, writing, or open-source contributions first on a resume. Ai2 offers internships and a predoctoral program that prepares people for graduate-level research.

Learning materials you can use now

• MIT OpenCourseWare is a free, open collection of MIT course materials with no enrollment or account required; MIT offers no credit or certification for it (about OCW (external site: ocw.mit.edu)). One example is 6.036 Introduction to Machine Learning (external site: ocw.mit.edu).
• Google's Machine Learning Crash Course (external site: developers.google.com) features animated videos, interactive visualizations, and hands-on practice exercises. Its prerequisites page (external site: developers.google.com) asks for comfort with basic algebra and statistics and solid programming skills, ideally in Python. No Google account is needed except for the Colab exercises, and Google offers badges, not formal certification (help (external site: support.google.com)).
• NIST's AI Resource Center (external site: airc.nist.gov) collects the AI Risk Management Framework, its Playbook, and a glossary, useful background for policy, safety, and evaluation work.

Reading a posting and using the Jobs page

Separate requirements from preferences. Postings often split them, as in Cerebras's required and nice-to-have lists. The three Anthropic postings cited here list a minimum of a bachelor's degree or an equivalent mix of education, training, and experience, and CoreWeave says applicants need not meet every item. Then note fixed conditions: on-site work, on-call rotations, lifting limits, state licenses, or security clearances. Apply on the employer's own site.

The Jobs page lists current positions at organizations in the catalog, collected from employers' public job boards, and links to the employer to apply. USASI does not employ or recruit for these organizations and collects no applications (methodology). You can search and filter by organization, location, department, workplace type, and employment type; each listing shows when it was last seen in the employer's feed, and organizations without an automated feed keep a link to their official careers page. Company records such as CoreWeave show open positions and the official careers link too.

Worked example: from a networking class to a data center

Sam is a fictional reader invented for this page. Sam is taking networking courses at a community college and wants to work on AI infrastructure.

1. Pick a role family. Sam reads the OOH page for network and computer systems administrators and notes that some employers ask for a certificate or associate's degree.
2. Find real postings. On the Jobs page, Sam searches "data center technician" and opens the CoreWeave posting.
3. Sort the requirements. Fixed: on-site work, on-call shifts, lifting. Flexible: hands-on or education-based experience with hardware, data centers, networking, or scripting.
4. Fill gaps. Sam practices Linux and Python and uses OpenCourseWare for background.
5. Check other routes. Sam searches apprenticeship.gov for registered programs nearby.
6. Apply. Sam applies through the employer's own careers page.

Different goals, such as research, would point to graduate study instead.

What you can do next

• Browse the Jobs page and read how it works in the Jobs methodology.
• Explore the chips and compute hub and the evaluation hub.
• Read Reading AI infrastructure and energy claims and How to read an AI evaluation.
• Look up accelerator, data center, and evaluation harness in the glossary.

Sources

All read on October 8, 2026. Job postings change or close; those below were open when read.

• U.S. Bureau of Labor Statistics, Occupational Outlook Handbook: Computer and Information Research Scientists (external site: bls.gov), Software Developers, Quality Assurance Analysts, and Testers (external site: bls.gov), Data Scientists (external site: bls.gov), Network and Computer Systems Administrators (external site: bls.gov), Computer Hardware Engineers (external site: bls.gov), Electrical and Electronics Engineers (external site: bls.gov), Electricians (external site: bls.gov), Lawyers (external site: bls.gov)
• U.S. Department of Labor: O*NET OnLine Data Scientists (external site: onetonline.org), Database Architects (external site: onetonline.org), and search for "machine learning engineer" (external site: onetonline.org); Apprenticeship.gov Career Seekers (external site: apprenticeship.gov) and AI in Registered Apprenticeship (external site: apprenticeship.gov)
• National Science Foundation: Advanced Technological Education (ATE) (external site: nsf.gov)
• NIST: AI Resource Center (external site: airc.nist.gov)
• MIT OpenCourseWare: About OCW (external site: ocw.mit.edu), 6.036 Introduction to Machine Learning (Fall 2020) (external site: ocw.mit.edu)
• Google: Machine Learning Crash Course (external site: developers.google.com), Prerequisites and prework (external site: developers.google.com), Help Center answers on accounts (external site: support.google.com) and certificates (external site: support.google.com)
• Anthropic: Careers (external site: anthropic.com); postings for Research Engineer, Post-Training Model Evaluations (external site: job-boards.greenhouse.io), Senior Safeguards Policy Lead, Cyber Harms (external site: job-boards.greenhouse.io), and Data Engineer (external site: job-boards.greenhouse.io)
• Ai2: Careers (external site: allenai.org); postings for Research Scientist, Robotics (external site: job-boards.greenhouse.io) and Lead Product Designer, AI for the Planet (external site: job-boards.greenhouse.io)
• Cerebras: postings for Senior ASIC RTL Design Engineer, AI Hardware (external site: jobs.ashbyhq.com) and Site Reliability Engineer - Ops & Automation (external site: jobs.ashbyhq.com)
• CoreWeave: Data Center Technician – Express Your Interest (external site: coreweave.com)
• Crusoe: Journeyman Electrician (external site: jobs.ashbyhq.com)