Session: How to deliver a machine learning project from data collection to production
Machine learning hype is fading out, but the technology is here to stay. Thousands of people are transitioning from data analysis, economics, finance, etc., to the data science field and honing their skills in programming and math. Tons of ML and MLOps frameworks have been built to support this fast growing field. But ML in practice still contains high risks, difficulties, and even when ML models are finally deployed into production, it still can be a failure (recall a Microsoft bot who quickly became a racist by chatting with people). Moreover, with new state-of-the-art architectures like GPT-3, fake news and spamming bots are becoming even bigger issues.
In this talk the speaker will share how data science teams can mitigate risks related to ML projects and deploy ML models successfully into production with low probability that something goes wrong in the production environment.
This recording is not available yet.
Bio: Sophia Aryan
Sophia is a co-founder of Bugout.dev and an open-source evangelist. As a former OpenAI scholar, she conducted research on the intersection of NLP and Reinforcement Learning.