Re-imagining distributed cloud database Amazon Aurora with focus on availability, resource management, serverless and database upgrades.

Held different roles as software development engineer 2 and SDE 1 at Amazon Aurora PostgreSQL engine team. Delivered more than 5 major features, lead more than 5 engineers building the world's most available cloud database at Amazon Web Services (AWS) relational database systems. https://aws.amazon.com/rds/aurora- Lead for Aurora PostgreSQL Serverless V2.- Improved cloud database debuggability- Lead for Aurora PostgreSQL availability and MAZ SLA compliance.Working on improving the availability, capacity, and performance of Amazon Aurora RDS.

● Performance Modeling: Applied supervised machine learning and statistical learning including regression techniques on features of an arbitrary parallel application for predicting runtime and performance. (2018)● Program Analysis based Performance Modeling: Devised plans, designed system architecture, and developed parallel application performance prediction framework (PyPassT) using static analysis of a program with an advanced compiler and HPC simulation of the application model. (05/2017 -- 03/2018)-->Published at ACM SIGSIM PADS 2018, May'18, Rome, Italy. [See publications]● Workload Simulator: Planned roadmap, designed system and developed capabilities to model parallel workloads using parallel HPC simulation. The simulator offers job scheduling, task mapping, and background traffic modeling for measuring runtime performance variability. (2016 -- 05/2017) --> Published at Wintersim WSC'17, Las Vegas, NV. [See publications]● Performance Predictuin Toolkitt: Developed HPC interconnection simulator as a detailed packet based MPI middleware in collaboration with Los Alamos National Laboratory LANL. (08/2015 -- 02/2016)--> The extended version of this research was published at ACM SIGSIM PADS 2016. [See publications]● SDN ScaleNet - Developed 5x efficient distributed real-time simulator for experimenting with distributed at-scale simulated and Software Defined Networks (SDN) featuring cloud systems. (08/2015 - 02/2017)--> This research work was published at SIMUTOOLS'17. [See publications]● PrimoGeni Constellation (PGC): Designed, implemented and deployed PrimoGENI Constellation (PGC) which is a parallel real-time simulation framework on NSF-GENI (Global Environment for Network Innovations) which provides a virtual laboratory for networking and distributed systems research. [See publications]. (04/2013 -- 05/2016)

Instructed and graded Programming III, Telecommunications Technology Applications and Spreadsheet Analysis.

Investigated high-performance interconnection networks in Cielo supercomputer (LANL, top rank 10 at top 500) and developed an early prototype of CRAY's Gemini torus interconnection performance model in Python using a parallel discrete-event simulator. The work had been matured to Performance Prediction Toolkit (PPT) which is in use for predicting parallel applications and HPC systems.--> Duration: May 11, 2015, to July 31, 2015

Designed courses, conducted lectures and exams on fundamental and advanced computer science classes to students. Courses: Parallel and Distributed Systems, Compiler Design, Object Oriented Programming C++, Fundamentals of Computing.