After several wonderful years at Brown University, I am going to join MIT in January 2018 as an Associate Professor.
I'm an Assistant Professor in the data management group of the Computer Science department at Brown University.
Currently my research focuses on building systems for machine learning or using machine learning for systems.
For example, with Vizdom we are exploring a new user interface to enable layman users to explore and build complex (ML) models, whereas with IDEA and Tupleware we develop new techniques to power the new workloads created by the next generation of user interfaces like Vizdom; the main challenge here is to to ensure interactive latencies regardless of the data size and type of operation. Similarly our work on auto-tuning for machine learning algorithms, MLbase & TuPAQ, or our work on quantifying the risk factors of missing data, the unknown unknowns, aim to help users to make faster and more sustainable discoveries.
Finally, with our work on Learned Indexes we started to explore how we can enhance or even replace core systems components using machine learning models and early results suggest, that we are able to achieve orders-of-magnitudes improvement over state-of-the-art techniques and sometimes are even able to change the complexity class of certain algorithms.
Current Research Interests
- Systems for Interactive Data Exploration
- Infrastructure for rack-scale analytics and machine learning
- Transaction processing over high-speed networks
- New consistency and concurrency control models
- Hybrid human-machine data management systems
In the following, a list of my current and past research projects:
- VizDom - Data Exploration on Interactive Whiteboards
- Tupleware - Redefining Modern Analytics on Modern Hardware
- QUDE - Quantifying the Uncertainty in Data Exploration
- MLBase - The Distributed Machine-Learning Management System
- S-Store - A streaming OLTP system for big velocity applications
- MDCC - The Fastest Strong Consistent Multi-Data Center Replication Protocol
- CrowdDB - Answering Queries with Crowdsourcing
- PIQL - Performance Insightful Query Language
- Cloudy/Smoky - a distributed storage and streaming service in the cloud
- Building a database on cloud infrastructure
- CloudBench - a benchmark for the cloud
- Zorba - a general purpose XQuery processor implementing in C++
- MXQuery - A lightweight, full-featured Java XQuery Engine
- Mapping Data to Queries (MDQ) - data integration with XQuery
- XQIB - XQuery In the Browser
Source: Geek & Poke - Cloud Comic