Back

Una Galyeva

Founder

@

Honest Data and AI Consulting

Session:

5 ways to fail scaling MLOps and Responsible AI within your organisation.

Interview with Una

Why do you want to speak at W+DAI?

Meet powerful women and learn from their experiences. Share own technical challenges and crowdsource potential solutions.

What is one person in your field of work that you admire and why?

Eventually there are many of them. If I need to choose one, I will go with Mehrnoosh Sameki - Responsible AI and AI Evaluation Tools Lead at Microsoft. Her contribution to several AI ethics open-source tools such as Fairlearn, Error Analysis, Responsible-AI-Toolbox and InterpretML.

Why would you recommend someone to join this field?

This is fantastic time to join and contribute to this highly-paced field with a lot of challenges and exciting new opportunities, where fresh perspectives and different backgrounds are needed.

What is the hottest topic/thing you have learned in the last few months, and why?

My topic does not sound like hot, however I guess it is one of the biggest pain points for engineering community. I was wrapping my head around the way to reduce the size of Docker images and their build time for some GenAI apps with self-hosted LLMs.

What is something funny that you have found out about the industry you are working in related to Data or AI?

It's more dark humour, though. The way some folks trying to apply LLMs for any scenarios even if they are not needed at all or costs and implications are not fully considered.

Why did you choose to give a talk about this topic specifically?

One set of MLOps talks covering mostly technical aspects, another high-level organisational ones. With this talk I aim to bridge the gap and provide holistic view on the MLOps challenges at hand.

What’s there to learn for our attendees?

Learn from some of the most common pitfalls experienced by organisations while scaling MLOps and implementing Responsible AI from technical and non-technical perspectives. Help to recognize the first signs of "MLOooops" happening at their organisation and equip them with powerful mitigations techniques.

Who is this talk for?

Data and AI engineers, AI product owners, Data and AI engineering managers, Heads of Data and AI and anybody else involved in MLOps and Responsible AI implementations.