Invest in the eval set
Without a stable, realistic evaluation set, changing models is guesswork.
INDUSTRY 02 · Artificial Intelligence
Artificial Intelligence
Model capability, data assets and inference cost, spanning compute through applications.
C
How many companies are in this industry
Illustrative data · describes structure, not official statisticsSee the size mix and where role demand is heading. Structural illustration — sources welcome.
Related companies~24,000 related companies
Algo & eng roles +26% YoY (illustrative)
Company-size distribution
Micro / <50
61%
Small / 50–200
24%
Medium / 200–1k
10%
Large / >1k
5%
Role demand mix
Algo & data
38%
Engineering & platform
40%
Product & solutions
22%
Value-chain position
- Upstream · compute & data
- Midstream · models & tooling
- Downstream · vertical apps
01
Trends
Trends
- From demos to stable business loops
- Latency and cost as product constraints
- Evaluation sets matter more than model choice
02
1 micro-notes
each with a read countG
Job Simulator
SimulateWalk through a few realistic situations and feel what this role or industry weighs every day.
AI · New Model or Better Eval
When results wobble, don't swap the model first.
D
Discussion
DiscussSpotted something inaccurate? Leave your experience here for everyone.
E
Salary is split by company size rather than one flat range, which makes comparison easier. Local ranges are welcome.
G
The interview questions are useful; a note on explaining your reasoning during live coding would help.
W
Propose an edit
EditLike a wiki, any field can be improved: pick a field, write the suggestion and your reason. Accepted proposals earn contribution points.
→
Common roles
→