Jan 2023 – May 2023 · 4 mos · Ithaca, New York, United States

• TA for the course Designing Data Products offered by Prof. Lutz Finger.

Apr 2021 – Aug 2022 · 1 yr 4 mos · Remote

• Pioneered the implementation of an Apache Parquet based storage format to enable customers to query Tempo through text matchers efficiently. Radically increased usability by increasing text based query speeds.
• Solved for poor query performance of key-value queries by remodeling the query path in Tempo using sharding and parallelism resulting in reduction of 95th percentile query latency from 5s to 2s.
• Added an end-to-end integration testing suite to thoroughly test writing, storage and querying of data, enhancing stability of Tempo.

Nov 2019 – Mar 2021 · 1 yr 4 mos · Remote

• Co-creator of Tempo - A high volume and cost effective distributed tracing backend.
• Added Amazon S3 support for long term data storage in Tempo - onboarded customers running applications on AWS.
• Performed profiling and benchmarking of the ingest path of Tempo and removed bottlenecks to scale ingest volume by 10x from 40MB/s to 400MB/s.
• Led the Production Readiness Review for GrafanaCloud hosted Tempo, involving an investigation of fault tolerance and recovery strategies. This was critical to release version 1.0 of Tempo and make the service Generally Available (GA).

Nov 2018 – Feb 2020 · 1 yr 3 mos

Jun 2018 – Nov 2019 · 1 yr 5 mos · Bangalore

May 2017 – Aug 2017 · 3 mos · WFH.

• Successfully completed Google Summer of Code 2017 under the LLVM Compiler Infrastructure.
• Researched and developed techniques to improve the performance of TensorFlow in training machine learning models, and noted a 30% improvement by leveraging LLVM and Polly.
• Presented my work at the LLVM Developer Conference, San Jose, CA. Link - https://www.youtube.com/watch?v=uq67__tfdtQ

Jun 2016 – Jul 2016 · 1 mo · Tokyo, Japan

• Recommended by Prof. Ramakrishna Upadrasta to a three week Research Internship program at the University of Tokyo. Worked on XFOR, a loop control structure that gives fine grained control of loop schedule to the programmer. Went on to learn about the cutting edge research at UTokyo including diamond tiling algorithms, GPU scheduling for stencil codes, etc. Brief work on ILP, Instruction Scheduling, Vectorization.

Aug 2014 – Jul 2018 · 3 yrs 11 mos · Hyderabad, India

• I took a keen interest in multiple fields of Computer Science including Distributed Systems, Compiler Engineering, Artificial Intelligence and Computer Networking, and have worked on multiple projects in these areas.
• My resume is available at - https://goo.gl/3ekmRQ