Roseline Polle

Bio

I am a senior AI auditor, looking for opportunities to conduct research in Mechanistic Interpretability. I worked as a Responsible AI auditor and researcher at Holistic AI for 3 years, performing a high number of early AI audits in various sectors, such as recruitment, facial recognition and financial services. I was also part of the UCL research team carrying out four pilots to improve empirical evidence around AI assurance during the 2020 CDEI programme of work on that subject. I participated in the development of material for two AI ethics courses for the Alan Turing Institute: Assessing and mitigating bias and discrimination in AI 1 & 2, as well as give live workshops and talks around these subjects, notably at the 2023 Devoxx conference in London. Before this, I used to develop computational models for the Clean Energy sector at Bloomberg New Energy Finance.

I have a strong mathematical and scientific background from my degree at Ecole Polytechnique (France) and Msc in Machine Learning at UCL.

Publications

Nigel Kingsman, Emre Kazim, Ali Chaudhry, Airlie Hilliard, Adriano Koshiyama, Roseline Polle, Giles Pavey & Umar Mohammed (2022). Public sector AI transparency standard: UK Government seeks to lead by example.. Discov Artif Intell. https://doi.org/10.1007/s44163-022-00018-4

Hilliard, Airlie and Kazim, Emre and Koshiyama, Adriano and Zannone, Sara and Trengove, Markus and Kingsman, Nigel and Polle, Roseline (2022). Regulating the Robots: NYC Mandates Bias Audits for Ai-Driven Employment Decisions.. SSRN. http://dx.doi.org/10.2139/ssrn.4083189

Polle, Roseline; Kazim, Emre; Carvalho, Graca; Koshiyama, Adriano; Inness, Catherine; Knight, Andrew; Gorski, Caroline; Barber, David; Lomas, Elizabeth; Yilmaz, Emine; Thompson, Gary; Ahamat, Ghazi; Pavey, Giles; Platts, Katie; Szpruch, Lukasz; Gregorovic, Miro; Rodrigues, Miguel; Ugwudike, Pamela; Nachev, Parashkev; Rai, Peter; Treleaven, Philip; Goebel, Randy (2021). Towards AI Standards: Thought-Leadership in Ai Legal, Ethical and Safety Specifications Through Experimentation. SSRN. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3935987

Emre Kazim, Adriano Soares Koshiyama, Airlie Hilliard and Roseline Polle (2021). Systematizing Audit in Algorithmic Recruitment.. Journal of Intelligence. https://doi.org/10.3390/jintelligence9030046

Other resources

Blog post for the Institute of the Future of Work (IFOW): Fair enough? Intersectionality in machine learning.
Talk for Devoxx 2023: Bias in Machine Learning at Devoxx 2023.
Open access course for the Alan Turing Institute: Assessing and Mitigating Bias and Discrimination in AI.[Github]
Open access course for the Alan Turing Institute: Assessing and mitigating bias and discrimination in AI: Beyond binary classification.[Github]

Creative project

Football Opera: Sonification of football data. Dash app turning data from a football match into music. Published alongside a written piece in the The Sideline magazine.

CV

[pdf]