Everything referenced in the chapter, grouped by how central it is, with a short note on what each one is for. Sources reviewed for the October 2026 revision. Hiring surveys, job-posting analyses and compensation submissions cover different populations, and none predicts an individual outcome. Check current postings and local pay before acting on an old figure.
Browse: Start here · Optional depth · Keep for reference
Start here
How to Interview and Hire ML/AI Engineers
Eugene Yan and Jason Liu · Practitioner account · Free · under an hour
A hiring-side rubric for turning project decisions into interview evidence. See Chapter 1's annotation. Confirm each employer's actual format with the recruiter.
AI Engineer Compensation Trends (Q3 2025)
Alina Kolesnikova, levels.fyi · Historical analysis · Free · under an hour
Useful for seeing how one pooled AI premium can hide differences by level. The site's 2025 sample reports a 6.2% AI/non-AI difference at entry level and 18.7% at staff level. These are historical patterns in submitted pay, not a causal return from studying AI or a current salary target. The article is now over a year old, so pair it with current local postings and the live tracker below. Self-selection and the employer and region mix are limitations.
normcore-llm-reads
Vicki Boykis · Curated reading list · Free · varies
A named engineer's reading list can help narrow a noisy feed. Pick one item that answers a question in your project, follow its evidence, and stop when you have the answer. A curated list is a starting point, and some links may have aged. Boykis's essay on building AI applications gives useful practitioner context on operating details that demos often omit.
State of AI Report 2026
Nathan Benaich and Air Street Capital · Annual industry report · Free · selected sections
The current edition covers research, industry, policy and safety. Use it to find a research direction or sector, then read the linked original paper or company disclosure and check real openings. It is investor-written analysis with investment interests, not an independent hiring study. A benchmark gain, funding announcement or revenue forecast does not show demand for junior staff. The site root rolls forward, so record the edition and slide number when citing a finding.
Optional depth
AI Exposure Isn't Squeezing Advertised Pay in the US — It's Boosting It
Jack Kennedy, Indeed Hiring Lab · Original job-posting analysis · Free · under an hour
A September 2026 analysis of US postings that advertise an annual salary. Occupations more exposed to AI show faster advertised-pay growth, but occupation and seniority mix affect the comparison. The seniority-specific pattern is suggestive, not conclusive. Read the methodology before interpreting a headline. Salary-advertising postings skew more senior than all postings, exposure is scored per occupation, and advertised pay is not realised pay. It does not isolate AI's causal effect or show entry-level pay in another country.
CIRR verified bootcamp outcomes
Council on Integrity in Results Reporting · Outcome reports · Free · under an hour
Use it to check a specific programme and cohort before you pay for training. Find the complete report and check its dates, graduation denominator, employment definition, follow-up window and audit status. An old cohort or a placeholder rate cannot back a current placement promise. Software-engineering bootcamp outcomes do not directly predict an AI/ML self-study search. A missing current report is missing evidence, not proof that every provider fabricates outcomes. One school's headline cannot tell you how long every search takes.
PwC 2026 Global AI Jobs Barometer
PwC research team · Original job-ad analysis · Free · selected sections
The report and its official release analyse over a billion job adverts across 27 countries/territories and separately examine US entry-level skill requirements. Use the definitions, period and comparison groups to see how advertised skills are changing. Its reported 62% average AI-skill wage premium is not a beginner's expected raise. Observational patterns cannot show that AI caused hiring or productivity changes. PwC sells AI advisory services too. Build a local role-requirement worksheet instead of copying a global premium into your salary target.
Machine Learning Interview Prep Guide
Aced, formerly Exponent · Commercial guide · Free article · selected sections
An optional checklist of interview topics and formats. Compare it with the hiring-side account above and the employer's instructions. Proprietary preparation frameworks are not industry standards. The company sells interview preparation, so separate the free topic map from its product recommendations. Tool permissions, round counts and expectations vary by employer and can change faster than a year-labelled article.
Keep for reference
AI Engineer Roadmap
roadmap.sh maintainers · Topic map · Free · lookup
A vocabulary and coverage cross-check, especially for application work with pretrained models. Its AI and Data Scientist map has a different scope. Use either to look up an unfamiliar term, then return to the chapter sequence and your project. A topic graph cannot tell you the time, prerequisite depth or evidence your role needs. Nodes differ in value, so check each linked resource's current version.
ML / AI Software Engineer salary tracker
levels.fyi · Submitted compensation tracker · Free lookup · varies
Filter by location, level, employer and recent date. Separate base salary, bonus and equity, because total pay can move with stock valuation. Record the sample and filters with any comparison. US aggregate medians do not describe an entry-level candidate in another region. Submitted data are self-selected and may under-represent smaller employers and lower-paid offers. Use several current local sources when negotiating.
ML-YouTube-Courses
DAIR.AI maintainers · Course index · Free · lookup
A handy way to find university lecture series. Before committing to a playlist, check the actual course site, year, assignments and software versions. Older fundamentals can still be useful, though an old framework walkthrough may need adapting. The index is a collection, not an ordered curriculum, and repository popularity does not prove learning outcomes. Use one course to fill a specific gap instead of adding several full courses.
Open Source Society University — Computer Science
OSSU contributors · Broad self-study curriculum · Free/varies by provider · long-term
A route to broader computing foundations for anyone who needs them. An experienced developer can pick one gap such as algorithms or systems instead of starting a full degree-sized curriculum. The materials do not lead to an accredited degree, and finishing coursework does not guarantee a job. Course availability, optional certificates and time needed depend on the providers.