Competitions: Beyond the Kaggle Leaderboard
In her third appearance on the DataTalks.Club Podcast, Tatiana Habruseva returns to talk about what has changed since her last interview, from becoming a mother and relocating to Silicon Valley to continuing her work in ML research, evaluation, and responsible AI.
We’ll cover two main themes:
Career growth through competitions: How platforms like Kaggle, AIcrowd, and Topcoder can help build a portfolio, create visibility, and open doors to research, consulting, and job opportunities, if used strategically.
Evaluation, benchmarks, and responsible AI: Why evaluating modern AI systems and agents is getting harder, why benchmarks need to evolve, and what responsible AI looks like in practice.
We’ll also discuss:
Why winning competitions is not enough on its own
How to turn technical work into papers, open source, and career opportunities
The pros and cons of ML competitions
Benchmark saturation, contamination, and alignment challenges
About the Speaker
Tatiana Habruseva, Ph.D. is a Staff Software Engineer, Machine Learning at LinkedIn, where she leads applied R&D on recommendation systems, multimodal alignment, and AI systems benchmarking. Before joining LinkedIn, she developed AI-based decision support systems for labor monitoring at Cork University Maternity Hospital. Her work on Weighted Boxes Fusion, a method for combining predictions from object detection models, has been cited over 700 times and ranks in the top 0.1% cited publications in computer science. Tatiana is a Kaggle Competitions Master and the 1st prize winner of the international Sound Demixing Challenge 2023. She holds a Ph.D. in Applied Physics from Cork Institute of Technology, is a Senior IEEE Member, and has authored over 27 peer-reviewed publications in computer science and physics.
DataTalks.Club is the place to talk about data. Join our Slack community!