JobPOVs
Explore

Applied AI systems

AI Prompt Engineer

A job title that may not exist in five years, folded into others that will.

Typical pay

$90k-$180k

How you get in

Technical background plus demonstrated evaluation work

Outlook

Uncertain as a title; the underlying skills are becoming standard

Prompt engineers design and evaluate the instructions and pipelines that get reliable output from AI models. The standalone title is already consolidating into broader AI product and engineering roles, and the durable skill is systematic evaluation, not clever wording.

A day in the life

  1. 9:00aReview failure cases from yesterday's evaluation run
  2. 10:00aRevise prompts and test against the benchmark set
  3. 12:30pLunch
  4. 1:30pBuild an evaluation harness for a new use case
  5. 3:00pWork with engineering on retrieval and context handling
  6. 4:30pDocument what works and why, so it survives model updates

How to get in

Engineer into applied AI

Duration

6-18 months

Cost

$0-$3,000

Credential

None — shipped AI features

  1. Work as a software engineer or data scientist first
  2. Build real applications on model APIs, including retrieval and tool use
  3. Learn systematic evaluation — this is the part that is actually scarce
  4. Move into an AI engineer or applied AI role

Domain expert into AI product

Duration

6-18 months

Cost

$0-$3,000

Credential

None — domain plus AI fluency

  1. Bring deep expertise in a field — law, medicine, education, finance
  2. Learn how models fail in your domain specifically
  3. Build evaluation sets that encode real expert judgment
  4. Domain knowledge plus AI fluency is more valuable than prompting alone

Treat it as a skill, not a job

Duration

3-6 months

Cost

$0-$1,000

Credential

None

  1. Add AI fluency to whatever role you already hold
  2. Automate real work in your current job and document the results
  3. Expect the standalone title to fold into engineering and product roles
  4. The durable version of this career is having another discipline underneath it

What people love

  • · Low formal barrier to entry right now
  • · Remote-friendly and well paid
  • · Genuinely new problems
  • · Skills fold into product, engineering and research roles

What wears people down

  • · The standalone title is already consolidating away
  • · Model updates can invalidate months of tuning
  • · Hype has flooded the market with weak candidates
  • · Real roles require engineering skill, not just prompt writing

From people doing it

Add yours