Alex Keller
theoretical computer scientist, Machine learning researchers, and Data Scientist
About Me
I’m Alex, a mathematician and data scientist, and member of House Bayesian.
My Interests
- Math
- Machine Learning
- Decision-making under uncertainty
- Bayesian optimization
- Probabilistic modeling
- Gaussian processes
- Active learning
- Uncertainty quantification
- Spatiotemporal modeling
With over a decade of experience as a head of data science and Lead Data Scientist, I have significantly impacted the projects I've participated in. My contributions have led to delivering impactful data science and machine learning end-to-end solutions and deploying machine learning models to production at organizations such as Google, Amazon, ING, and Nike.
During my recent tenure, I focused on the fintech sector, particularly at Tillfull, a startup where Nav Technologies successfully acquired our technology. Faced with small amounts of labeled data and high-risk decision spaces, we designed a platform that leverages weak supervision and generative models to create a probabilistic label set that is then passed to various downstream models. Please see (see https://arxiv.org/abs/2305.18430).
My expertise spans various technical areas, including adaptive experimentation (Bayesian optimization, multi-armed bandit optimization), Graph processing (vertex and edge centric), graph representation learning, graph transactional databases, everything Bayesian, DNN, recommendation systems, search ranking and relevance, online learning, weak supervision, and natural language processing (NLP. Q/A, NER, Fine-tuning, Embedding construction).
AI and machine learning have made remarkable strides, but a significant challenge remains: effectively handling uncertainty. Many models struggle to express their confidence level and make decisions that account for ambiguity. This shortcoming limits AI deployment in crucial areas like Healthcare, Law, and Finance, where quantifying uncertainty is vital, and in dynamic real-world settings where models may encounter unfamiliar data. I'm now taking on these challenges in the healthcare sector!