01Roles, skills and tasks / Tasks

What the work actually is.

The tasks 1,937 AI job ads since 1 October ask you to do, which role asks for each one most, and the daily work of every role.

the ads ask you to21% Lead and mentor engineers

Most of all Agentic AI Engineer: 25% of their ads.

Every task in the ads

Share of AI job ads that ask for each task.

  • Lead and mentor engineers21% · 352
  • Build agents and tool use21% · 349
  • Ship AI features to production20% · 331
  • Prototype and iterate fast20% · 329
  • Integrate with existing systems17% · 284
  • Train and fine-tune models15% · 257
  • Evaluate models and outputs14% · 234
  • Build data pipelines and datasets13% · 210
  • Work with customers11% · 188
  • Scale and optimise infrastructure11% · 176
  • Build retrieval and search5% · 89
  • Research new methods4% · 67
Role by task

Which roles do which work.

Read across a row to see what a role spends its days on, down a column to see who does a task most. Hover a cell for the number.

Lead and mentor engineersBuild agents and tool useShip AI features to productionPrototype and iterate fastIntegrate with existing systemsTrain and fine-tune modelsEvaluate models and outputsBuild data pipelines and datasetsWork with customersScale and optimise infrastructureBuild retrieval and searchResearch new methods
Applied AI Researcher363Applied AI Researcher · Lead and mentor engineers: 15%Applied AI Researcher · Build agents and tool use: 10%Applied AI Researcher · Ship AI features to production: 12%Applied AI Researcher · Prototype and iterate fast: 20%Applied AI Researcher · Integrate with existing systems: 7%Applied AI Researcher · Train and fine-tune models: 25%Applied AI Researcher · Evaluate models and outputs: 19%Applied AI Researcher · Build data pipelines and datasets: 17%Applied AI Researcher · Work with customers: 3%Applied AI Researcher · Scale and optimise infrastructure: 10%Applied AI Researcher · Build retrieval and search: 2%Applied AI Researcher · Research new methods: 13%
AI Platform Engineer333AI Platform Engineer · Lead and mentor engineers: 20%AI Platform Engineer · Build agents and tool use: 13%AI Platform Engineer · Ship AI features to production: 16%AI Platform Engineer · Prototype and iterate fast: 13%AI Platform Engineer · Integrate with existing systems: 13%AI Platform Engineer · Train and fine-tune models: 16%AI Platform Engineer · Evaluate models and outputs: 12%AI Platform Engineer · Build data pipelines and datasets: 11%AI Platform Engineer · Work with customers: 2%AI Platform Engineer · Scale and optimise infrastructure: 14%AI Platform Engineer · Build retrieval and search: 5%AI Platform Engineer · Research new methods: 2%
Forward Deployed Engineer260Forward Deployed Engineer · Lead and mentor engineers: 20%Forward Deployed Engineer · Build agents and tool use: 22%Forward Deployed Engineer · Ship AI features to production: 26%Forward Deployed Engineer · Prototype and iterate fast: 19%Forward Deployed Engineer · Integrate with existing systems: 25%Forward Deployed Engineer · Train and fine-tune models: 3%Forward Deployed Engineer · Evaluate models and outputs: 3%Forward Deployed Engineer · Build data pipelines and datasets: 17%Forward Deployed Engineer · Work with customers: 33%Forward Deployed Engineer · Scale and optimise infrastructure: 2%Forward Deployed Engineer · Build retrieval and search: 5%Forward Deployed Engineer · Research new methods: 0%
Agentic AI Engineer206Agentic AI Engineer · Lead and mentor engineers: 25%Agentic AI Engineer · Build agents and tool use: 41%Agentic AI Engineer · Ship AI features to production: 19%Agentic AI Engineer · Prototype and iterate fast: 12%Agentic AI Engineer · Integrate with existing systems: 15%Agentic AI Engineer · Train and fine-tune models: 12%Agentic AI Engineer · Evaluate models and outputs: 16%Agentic AI Engineer · Build data pipelines and datasets: 7%Agentic AI Engineer · Work with customers: 6%Agentic AI Engineer · Scale and optimise infrastructure: 12%Agentic AI Engineer · Build retrieval and search: 2%Agentic AI Engineer · Research new methods: 2%
Applied AI Engineer197Applied AI Engineer · Lead and mentor engineers: 16%Applied AI Engineer · Build agents and tool use: 20%Applied AI Engineer · Ship AI features to production: 23%Applied AI Engineer · Prototype and iterate fast: 23%Applied AI Engineer · Integrate with existing systems: 12%Applied AI Engineer · Train and fine-tune models: 16%Applied AI Engineer · Evaluate models and outputs: 22%Applied AI Engineer · Build data pipelines and datasets: 8%Applied AI Engineer · Work with customers: 9%Applied AI Engineer · Scale and optimise infrastructure: 1%Applied AI Engineer · Build retrieval and search: 10%Applied AI Engineer · Research new methods: 1%
AI-Native Developer168AI-Native Developer · Lead and mentor engineers: 15%AI-Native Developer · Build agents and tool use: 18%AI-Native Developer · Ship AI features to production: 22%AI-Native Developer · Prototype and iterate fast: 10%AI-Native Developer · Integrate with existing systems: 21%AI-Native Developer · Train and fine-tune models: 7%AI-Native Developer · Evaluate models and outputs: 5%AI-Native Developer · Build data pipelines and datasets: 8%AI-Native Developer · Work with customers: 8%AI-Native Developer · Scale and optimise infrastructure: 11%AI-Native Developer · Build retrieval and search: 5%AI-Native Developer · Research new methods: 0%
AI Solutions Architect85AI Solutions Architect · Lead and mentor engineers: 25%AI Solutions Architect · Build agents and tool use: 16%AI Solutions Architect · Ship AI features to production: 12%AI Solutions Architect · Prototype and iterate fast: 29%AI Solutions Architect · Integrate with existing systems: 11%AI Solutions Architect · Train and fine-tune models: 18%AI Solutions Architect · Evaluate models and outputs: 1%AI Solutions Architect · Build data pipelines and datasets: 11%AI Solutions Architect · Work with customers: 22%AI Solutions Architect · Scale and optimise infrastructure: 15%AI Solutions Architect · Build retrieval and search: 9%AI Solutions Architect · Research new methods: 0%
AI-Native Architect58AI-Native Architect · Lead and mentor engineers: 24%AI-Native Architect · Build agents and tool use: 22%AI-Native Architect · Ship AI features to production: 9%AI-Native Architect · Prototype and iterate fast: 28%AI-Native Architect · Integrate with existing systems: 24%AI-Native Architect · Train and fine-tune models: 10%AI-Native Architect · Evaluate models and outputs: 21%AI-Native Architect · Build data pipelines and datasets: 9%AI-Native Architect · Work with customers: 34%AI-Native Architect · Scale and optimise infrastructure: 3%AI-Native Architect · Build retrieval and search: 2%AI-Native Architect · Research new methods: 3%
0%100%share of that role's ads; numbers shown from 50%
Tasks

What each role does all day.

Build

AI-Native Developer.

  • Turn a vague goal into a shipped feature
  • Drive changes with coding agents and custom skills
  • Review AI-written code critically before it merges
  • Add AI features to the product and measure them
Build

Applied AI Engineer.

  • Frame the problem and find the data
  • Build retrieval and prompts that hold up
  • Write evals from real tasks and gate releases on them
  • Control cost and latency in production
Build

Agentic AI Engineer.

  • Design execution loops and tool interfaces
  • Orchestrate multiple agents over MCP
  • Make long-running tasks resumable and safe to fail
  • Trace and analyse agent failures
Deploy

Forward Deployed Engineer.

  • Run discovery inside the customer's team
  • Prototype fast, then harden for production
  • Integrate with the customer's data and systems
  • Train teams, drive adoption, measure ROI
Design

AI Solutions Architect.

  • Map the customer's systems and data
  • Decide build versus buy
  • Design for governance and compliance
  • Write the reference architecture
Design

AI-Native Architect.

  • Write the spec and the architecture
  • Set the context and token strategy
  • Trade off cost, latency and reliability
  • Define how the system is evaluated and governed
Platform

AI Platform Engineer.

  • Run the model gateway and routing
  • Serve and cache models efficiently
  • Attribute cost to teams and features
  • Keep tracing and evals running for everyone
Research

Applied AI Researcher.

  • Design benchmarks and ablations
  • Optimise prompts and pipelines against a metric
  • Fine-tune or adapt models when it pays
  • Publish and turn results into practice

Certificates and degrees have their own page: Certificates.

More in roles and skills