Distribution across 38 profiles.
Middle half of AI Engineers score between 36% and 45%.
0%
50%
100%
p10 · 33%
49% · p90
Task breakdown by work type
On-screen work77%
Done entirely on a computer. High AI exposure — these tasks are already in the automation zone.
In-person + screen23%
Physical sensing, digital output — e.g. interviewing someone then writing a report. Partially protected.
Computer + action0%
Computer input, real-world output — needs someone to act on it, not just software.
Fully in-person0%
No computer required. Furthest from automation — the strongest human advantage.
Typical tasks
3 synthetic profiles for a AI Engineer, ordered by automation exposure.
Tab between them to see how task mix drives the score difference.
TaskTimeTypeExposure
Preprocessing and cleaning large datasets (e.g., text, images, tabular data) to prepare them for training, including handling missing values, normalizing features, and augmenting data to improve model robustness.
24%DD
65%
Collaborating with product managers, data scientists, and software engineers to define requirements, align model outputs with business goals, and integrate AI solutions into existing systems or products.
deep expertisesocial core
24%AD
16%
Designing and implementing machine learning models (e.g., neural networks, transformers) to solve specific business or research problems, including selecting architectures, tuning hyperparameters, and optimizing performance metrics like accuracy or latency.
deep expertisesocial element
15%DD
22%
Researching and experimenting with new AI techniques, tools, or frameworks (e.g., reading papers, attending conferences, or prototyping novel approaches) to stay current and improve existing solutions.
deep expertisesocial element
13%AD
14%
Monitoring and maintaining deployed models in production, including tracking performance drift, retraining models with new data, and troubleshooting issues like latency or prediction errors.
deep expertisesocial element
11%DD
26%
Writing, debugging, and optimizing code (e.g., Python, TensorFlow, PyTorch) to train, evaluate, and deploy models, including setting up pipelines for experimentation and version control (e.g., Git, MLflow).
9%DD
66%
Documenting workflows, model architectures, and decision-making processes for internal teams or stakeholders, including creating reports, presentations, or technical specifications.
1%DD
43%
TaskTimeTypeExposure
Designing and implementing machine learning models (e.g., neural networks, transformers) to solve specific business or research problems, including selecting architectures, tuning hyperparameters, and optimizing performance metrics like accuracy or latency.
deep expertisesocial element
25%DD
26%
Writing, debugging, and optimizing code (e.g., Python, TensorFlow, PyTorch) to train, evaluate, and deploy models, including setting up pipelines for experimentation and version control (e.g., Git, MLflow).
19%DD
94%
Researching and experimenting with new AI techniques, tools, or frameworks (e.g., reading papers, attending conferences, or prototyping novel approaches) to stay current and improve existing solutions.
19%AD
31%
Preprocessing and cleaning large datasets (e.g., text, images, tabular data) to prepare them for training, including handling missing values, normalizing features, and augmenting data to improve model robustness.
14%DD
66%
Monitoring and maintaining deployed models in production, including tracking performance drift, retraining models with new data, and troubleshooting issues like latency or prediction errors.
8%DD
57%
Collaborating with product managers, data scientists, and software engineers to define requirements, align model outputs with business goals, and integrate AI solutions into existing systems or products.
deep expertisesocial core
8%AD
7%
Documenting workflows, model architectures, and decision-making processes for internal teams or stakeholders, including creating reports, presentations, or technical specifications.
4%DD
48%
TaskTimeTypeExposure
Designing and implementing machine learning models (e.g., neural networks, transformers) to solve specific business or research problems, including selecting architectures, tuning hyperparameters, and optimizing performance metrics like accuracy or latency.
35%DD
53%
Preprocessing and cleaning large datasets (e.g., text, images, tabular data) to prepare them for training, including handling missing values, normalizing features, and augmenting data to improve model robustness.
16%DD
65%
Writing, debugging, and optimizing code (e.g., Python, TensorFlow, PyTorch) to train, evaluate, and deploy models, including setting up pipelines for experimentation and version control (e.g., Git, MLflow).
15%DD
74%
Documenting workflows, model architectures, and decision-making processes for internal teams or stakeholders, including creating reports, presentations, or technical specifications.
11%DD
56%
Monitoring and maintaining deployed models in production, including tracking performance drift, retraining models with new data, and troubleshooting issues like latency or prediction errors.
10%DD
51%
Collaborating with product managers, data scientists, and software engineers to define requirements, align model outputs with business goals, and integrate AI solutions into existing systems or products.
deep expertisesocial core
8%AD
8%
Researching and experimenting with new AI techniques, tools, or frameworks (e.g., reading papers, attending conferences, or prototyping novel approaches) to stay current and improve existing solutions.
deep expertisesocial element
2%AD
16%
Save & share
AI tools for this role
Tools relevant to the most automatable tasks in this profession.
