Sidhavratha Kumar

Application Development

• Started as Lead Engineer with small team of 2 engineers. Crafted and developed product from scratch.
• Evaluated emerging technologies to build systems at lightning speed.
• Worked on designing and building Order Management System from scratch.
• Built Carrier aggregator system, which is functioning over 7 countries.
• Build extensive geo-mapping system to support selection across multiple carrier.
• Designed and streamlined data-warehousing pipeline using BigQuery.
• Tools: Django, Python, Celery, PostgreSQL, BigQuery, GIS

Team Management:

• Built strong teams of 10 Backend, 5 Frontend, 2 Data engineers through selective recruitment and post-hire mentorship of junior engineers.
• Worked through multiple iterations of performance review process.
• Received highest MRX across company for 3 straight years.

AI and Machine Learning:

• Designed and helped team develop AI 1st Chatbot agent to help with system query and support (using LangChain and OpenAI).
• Developed ML pipeline using BigQuery and VertexAI for following use-case:
  - Package prediction model - Increase accurate order's package dimension prediction by 70%, which closed down the gap between estimated and actual charge of shipments
  - Inventory Product dimension accuracy - Help merchants find discrepancy in dimensions of their products, which could avoid mischarging storage or shipping fees
  - Forecast of stock depletion - 30% increase in OOS prevention for merchants
  - Transit time prediction - with accuracy for 80% based on different geography, and carriers performance
• Tools: LangChain, OpenAI, VertexAI, Gemini.

• - Managing team of 7 developers as Tech Lead in one of the most critical component of eCommerce site
• - Migrated microservice application to Kubernetes with HPA (horizontal-pod-autoscalar), which reduced cost by 30%
• - Eliminated Kafka data loss by creating Kafka lag monitoring tool which monitor/alerts based on time based SLA instead of count of messages
• - Created solution for business stakeholders using ElasticSearch / Kibana to provide on-the-spot rule evaluations
• - Helped team scale up applications by 4x during traffic surge due to COVID-19 pandemic scare
• - Helped team deliver multiple business critical features on-time
• Application: PnO (Price & Offer) Walmart Canada
• Tools: SpringBoot, Kafka, Apache Spark, Spark Streaming, Cassandra, Micro-Services, Multi-Zone deployments

• - Increased cache hit ratio of micro-services to 98% from 70%, which contributed to reduction of response time by 20%
• - Applied concurrency-limits library to protect nodes from high load, and implement fail-fast
• - Optimized application code to reduce CPU utilization by 10%
• - Added custom Kafka related fix to speed performance of Spark Job by 30%
• - Fixed many spark jobs, some of which shown improvement of 700%
• - Created E2E test environment setup using docker and docker-compose
• - E2E test execution time reduced to 15 minutes from 2-5 days
• - System is designed as HA to fail and recover without customer impact
• - Re-designed/developed/deployed and managed critical pricing engine for e-commerce site for Walmart
• Application: PnO (Price & Offer) Walmart Canada
• Specialized on: Cassandra/Kafka/Spark Streaming/HDFS, Springboot with REST/Microservices, Docker & Kubernetes
• Tools: Springboot, Kafka, Apache Spark, Spark Streaming, Cassandra, Microservices, Multi-Zone deployments

• - Designed and re-written legacy reports (ex
• Balance-sheet, Regulatory reports, ...) as Spark calculation and integrated with Spark/HDFS/YARN cluster to generate the reports as scheduled jobs
• - Balance-sheet report generation uses 100 GB of data which gets processed within 2 mins and results are available to user for review/action
• - Implemented multiple internal optimization to achieve super-fast response time for huge amount of data
• Application: Ignite
• Tools: Apache Spark, Hadoop MR, HDFS, YARN, Sqoop, AVRO, Java 8

• - Designed and developed Ignite application which can serve as platform for calculation/jobs written for Apache Spark
• - Platform enables data to be sourced and written to different endpoints (for ex
• REST, Database, HDFS, Local File, ...)
• - Platform enables other application developers to decouple their logic into source, calculation and target, which can be managed through configuration
• - Implemented integration test for application, which can be imported as library and executed as separate instance in JUnit tests
• - Debugged and resolved various authentication and configuration related issues with YARN and HDFS
• - Designed error catalog for application to increase transparency of application errors
• - Developed testing tools and ecosystem, by which override on Apache Spark / HDFS library level code changes could be tested with application
• Application: Ignite
• Tools: Apache Spark, HDFS, YARN, REST, Hystrix, Java 8

• - Designed and developed XRS (eXternalised Result Set) application which can store huge amount of data at very fast speed (100 GB in 2 minutes) into high available data storage
• - Developed functionality in XRS to query (aggregate/filter/variance) stored data with split second response time
• - Developed security to control data access at column + row level based on user permissions
• - This application was used to store output of big data tasks, and used directly by reporting tools to view large set of data on real-time, using columnar database
• Application: XRS (eXternalised Result Set)
• Tools: Sybase IQ, REST, Java 8, Multithreading

• - Developed components of cache which enables distributed cache connected to milestoned tables in database, which refresh automatically as new data arrives
• - Used Off-Heap memory to store/manage data, to optimize over garbage collection
• - Developed various tools within cache to visualize and debug data based on dynamic logic or query
• - Cache can be connected to database hosted on various regions and provides localized view of data as if all data and their relation is loaded in one machine
• Application: Mithra
• Tools: Sybase ASE, Core Java, Multithreading, Java Off-Heap, REST, HTML, JQuery

Company Overview:

Project Company: VISA

• Created Merchant onboarding application which would ensure merchants are technically compliant before onboarding onto VISA
• Worked on captcha page which comes in-between 3rd Secure validation of a transaction
• Learned principals and internals of security in detail
• Created an internal application knowledge test tool-base
• Project Company: VISA
• Application: Verified By VISA (3rd Secure Page)
• Tools: Java, MySQL, Spring MVC, Hibernate, JQuery, HTML, CSS, Certificates, SSL

• Company Overview: Project Company: CitiBank
• - Joined team with 100+ developer and quickly learned the application and different tools
• - Built dynamic page by combining backend and frontend, which could display questionnaires to CS Agents based on different situations, with internally dependent components
• - Build framework to rollback system to any previous state of workflow by using Hibernate
• - Learned core of OS, Java, J2EE, and internals on how everything works together
• Project Company: CitiBank
• Application: Sawgrass (Customer Support Application)
• Tools: Java, Servlet, Chordiant, Hibernate, Oracle 11G, Javascript + JSF, OS, Socker, Networking