Infrastructure as code:- Implemented new architecture, extended existing configurations for network, and runtime environments (Terraform, AWS, ECS, Kubernetes, RDS)- Architected and implemented Faros platform on Kubernetes for on-premises deployments (Helm, AWS, Kubernetes)- Architected and implemented secure deployment pipeline for Kubernetes services (Buildkite, ECS)- Designed and implemented tools for security-compliant tenant and user management (Python, buildkite, AWS ECS)Performance, Reliability and Security:- Implemented test automation framework (Buildkite, Pytest, Sentry)- Developed performance, integration, and end-to-end tests (Buildkite, Github actions, Pytest, Selenium)- Instrumented services for continuous monitoring and alerting (Datadog, PagerDuty, Typescript, Scala)- Evaluated performance, prototyped, analyzed and compared cutting edge DB technologies (Aurora, Clickhouse, DuckDB)- Designed and implemented service-to-service authentication using JWTsOpen source contributions:- Airbyte https://github.com/airbytehq/airbyte/pull/17713, https://github.com/airbytehq/airbyte-platform/pull/349- Bitnami MLFlow https://github.com/bitnami/charts/pull/20582, https://github.com/bitnami/charts/pull/20879

Designed and implemented a large scale (few thousands orgs and ML prediction flows) performance test. Using this test, certified Einstein Platform for freemium launchAdopted test automation and platform’s UI portal to kubernetes platformSince April 2020, with the expansion of Einstein org, oversaw performance work for all Einstein products, including Chatbots, Search, ML services and applicationsDefined performance testing strategy in public cloud for newly developed Search as a Service platformEvaluated impact of latency with multi AZ implementation of services in public cloud by designing, implementing, running experiments, and analyzing their resultsLed availability initiative by defining SLOs/SLIs and collaborating with SRE team on implementation of SLOs/SLIs for all Einstein services and applicationsWith Einstein platform applications and services transitioning to Hyperforce (public cloud), defined performance testing strategy, implemented performance tests and tools for testing in the cloud.

Designed and implemented UI portal surfacing functionality and metadata of all platform API services. This tool significantly improved productivity of developers, product managers as well as enabled efficient operation of production and testing environments. (Python, Django, JQuery, Docker, Mesos/DCOS)Designed and implemented test automation framework which enabled rapid development of integration/performance/ML tests, user friendly test management and execution, automated test results collection, processing, and visualization (Python, Django, Google charts, S3, Git, JMeter)Designed and implemented test data generators for performance testing of ML workflows. Object records generated randomly or following a user specified regression model (Python, Scikit).Designed and implemented performance regression tests for ML training/scoring algorithms.

- First Fitbit's Performance Engineer and hence 360° view of performance problems: - design and implementation of end-to-end performance test automation which includes test orchestration, test execution, results and log collection, web service for test management and results visualization (Python+CherryPy, JMeter, Graphite) - prototyping new algorithms and systems (e.g., blocking vs non-blocking Kafka clients, unique ID generation in MySQL, Zookeeper, etc.) - designing and implementing performance tests for new systems and algorithms - troubleshooting production issues, JVM tuning in production - A/B testing in production, analysis of test results - capacity and scalability of tested and production systems - designed and implemented integration of the performance testing automation system with AWS to simulate large scale (production) load. This feature was used in testing Fitbit's social feed feature in dark mode in production to ensure successful day one launch to all customers.

- Filed application for US patent “System and Method for Performance Tuning of Garbage Collection Algorithms” ( a discrete-even simulation modeling of Java heap).- Key member of release SWAT team investigating performance regressions in production. Root cause analysis of the major incident in Spring'14 release. - Design, implementation and analysis of complex performance workload targeting transient behavior of the caching layer to simulate a major production incident. The workload recreates a state of the system when invalidation of cache objects and their subsequent rebuilding leads to failure of network, cache, and DB tiers.- Technical lead for all Apex related performance issues and workloads.- PoC of Extrahop network monitoring appliance, verification of its data against statistics reported by internal tools.- Published blog on solving Java memory regressions: https://developer.salesforce.com/blogs/engineering/2014/12/solving-java-memory-regressions-high-accuracy-zero-overhead.html- Architected performance tests, performance analysis of the newly introduced multi-tier (microservices) architecture.

- Performance testing, scalability, cost analysis and optimization of an incubator project on Internet of things (IoT), simulating millions of connections. - Designed load tests, orchestration of the tests in AWS using EC2 API, used Sumo logic for log analysis- Presented results of the IoT project at a session in Dreamforce 2013- Completed with distinction course Functional Programming Principles in Scala https://www.coursera.org/course/progfun- Performance testing, analysis of key components of Salesforce.com platform.- Post release performance monitoring and analysis, RCA of regressions. - Large scale (up to 1 million concurrent connections) performance testing of load balancer (F5).- Mentoring new hires.

Post release production performance analysis, RCA of production regressionsPOCs of performance monitoring tools (AppDynamics, dynaTrace, OPNET AppInternals).Performance testing, data analysis, and regression analysis of key platform components. Performance testing of new features. Enhancement of test automation tools.Analysis and simulation of Java heap and garbage collections.

Daily runs and analysis of workloads for performance testing of key platform components. Test data analysis. Root cause analysis of regressions using JMeter, Splunk, custom log parsers, statistical tools, Java profiliers (JProfiler, YourKit). Development of test automation tools. Evaluation of network monitoring tools and interaction with vendors of the tools.

Research and consulting in computer networks, network traffic, protocol simulation modeling and performance analysis, expert and advanced user of OPNET Modeler.

Manage computing facilities of the School of Computing and Engineering that include about 10 Windows/Linux servers and more than 300 personal computers, including 150 workstation in 6 computer labs. Supervise 4 technical assistants and 10 lab assistants.