Krishna Puttaswamy Email & Phone Number

San Francisco, Ca, Us

Remote First, Us

San Francisco, California, Us

San Francisco, California, Us

San Francisco, California, Us

I'm an engineer in Uber Marketplace (https://marketplace.uber.com/), the team responsible for solving some of the most exciting and challenging problems in Uber -- pricing trips, matching drivers and riders, incentive allocation and optimization, driver positioning, forecasting, etc.Ping me if these problems sound interesting to you!

San Francisco, Ca, Us

I lead the ML Infra team at Airbnb, where we are building an end-to-end platform, called Bighead, to support a variety of ML use cases at Airbnb. https://databricks.com/session/bighead-airbnbs-end-to-end-machine-learning-platform

San Francisco, Ca, Us

As an early engineer on the Data Infra team, I have had the opportunity to build many of the basic infrastructure for data processing at Airbnb. This infra has scaled with the rapid growth in data and data products.* I lead the project to rebuild our hadoop clusters. This lead to 3X improved performance at 1/4th the cost, while making it easier to incrementally scale to handle much more data. https://medium.com/airbnb-engineering/data-infrastructure-at-airbnb-8adfb34f169c#.h6axui493* I lead the project to revamp our data ingestion pipeline to reliably land the events generated on the site into hadoop/hive warehouse. The new system is super reliable and timely, with negligible amount of loss in events. See https://www.slideshare.net/HadoopSummit/reliable-and-scalable-data-ingestion-at-airbnb-63920989 and https://www.slideshare.net/alexismidon/jitney-kafka-at-airbnb for details* Built a scalable data analytics service for interactive slicing and dicing of data.* Built and productionized a K/V store for serving derived data. https://medium.com/airbnb-engineering/nebula-as-a-storage-platform-to-build-airbnbs-search-backends-ecc577b05f06

Sunnyvale, Ca, Us

Our team built a new data processing framework on Hadoop to efficiently perform Joins and Aggregations on large datasets. This framework, called Cubert, has proven to be significantly faster than open-source alternatives, such as Pig and Hive, for such aggregate computations. These improvement are due to novel data organization primitives, and memory-conscious implementation of the system — we have designed new data structures and algorithms to take full advantage of the speed of in-memory computation. For details, see the VLDB 2014 paper on this work (www.vldb.org/pvldb/vol7/p1462-vemuri.pdf). The framework is open sourced and available at https://github.com/linkedin/CubertI was a part of the team that built Espresso — LinkedIn’s distributed data serving platform (http://www.slideshare.net/amywtang/espresso-20952131). Lead our efforts on improving its performance: redesigned/reimplemented some performance critical components, diagnosed and fixed various concurrency bugs, and also built a service for capacity estimation while on boarding new clients.

My main focus in the recent past has been on: a) Optimizing the cost of running applications -- specifically storage applications -- on the cloud, and b) Improving the security of software-defined networks (SDNs), OpenFlow in particular, in data center environments.Selected publications related to these projects: "Frugal Storage for Cloud File Systems" Published at the Proceedings of ACM SIGOPS/EuroSys European Conference on Computer Systems (EuroSys 2012) "Lowering Inter-Datacenter Bandwidth Costs via Bulk Data Scheduling" [Best Paper Award] Published at the Proceedings of IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid 2012) Ottawa, Canada, May 2012.

Santa Barbara, Ca, Us

My PhD research focus was on distributed systems. One of the key contribution of my thesis was the design and implemented of a system called Silverline, which enables users to run their web applications on the cloud while keeping (most) data in encrypted form on untrusted clouds.For more details, refer to: "Silverline: Toward Data Confidentiality in Storage-Intensive Cloud Applications" published at The 2nd ACM Symposium on Cloud Computing (SoCC) 2011 Cascais, Portugal, October 2011.

Developed a system to edit Microsoft Office documents from mobile devices. See the following paper for more details."Docx2Go: Collaborative Editing of Fidelity Reduced Documents on Mobile Devices" Published at the 8th Annual International Conference on Mobile Systems, Applications and Services (MobiSys 2010) San Francisco, CA, USA. June 2010

Redmond, Washington, Us

Developed systems to a) automatically diagnose and fix network problems, and b) collect data from clients while preserving their privacy.See these papers for more details:"Deja Vu: Fingerprinting Network Problems" Published at the 7th International Conference on emerging Networking EXperiments and Technologies (CoNEXT) 2011 Tokyo, Japan, December 2011. "Anonygator: Privacy and Integrity Preserving Data Aggregation" Published at the 11th International Middleware Conference (Middleware 2010) Bangalore, India. November 2010

I worked as a developer on Visual Studio 2005 Build and Source Code Control Systems.

Developed tools to process large-scale financial transaction datasets.

Ph.D, Distributed Systems, Security And Privacy

Bachelor Of Engineering, Computer Science