Our services

What we do in a data project

1. Find it:

Consultation about the details of your data needs. We use bespoke bias metrics to judge the quality of your raw and synthetic data. We create reports about the representation and fairness of your data. We flag for potential concerns with using the data and how to mitigate those. 

2. Fix it:

We deploy bespoke algorithms and technical support that fix existing bias using machine learning processes both in pre-processed data and during the Machine Learning process. 

3. Use it:

We offer bespoke post-processing AI alignment discussions and documentation, including suggestions on how concepts of human-in-the-loop and transparency/explainability could be employed, how to incorporate reminders of where the synthetic data is more/less robust from a bias, fairness perspective, and how generalizable the synthetic data may or may not be. We can also come with bespoke suggestions for the next round of data collection, synthesis, and use from a bias, fairness, alignment perspective.

4. Teach your team:

Custom designed workshops on:

1. Fairness and bias, combining social science and ML elements
2. Finding bias in synthetic data – how to do it
3. Bias in your team - How to hire and retain a diverse team
4. Tips on dealing with uncertain, qualitative data

We would like to help.