Dec 2024 – Aug 2025 · 8 mos

• Built low latency real time quantitative analysis pipeline for crypto trading.

May 2022 – Jun 2024 · 2 yrs 1 mo · Pune, Maharashtra, India

• SxT's sub-second, zk-proven, cryptographically guaranteed, real-time onchain/offchain data analysis makes smart-contracts smarter.
• Decentralized data warehouse accessible by smart contracts for real-time analysis to make data-driven contract execution.
• Part of data warehouse team to build a HTAP database on top of Apache Ignite using Apache Spark.
• Lead a team of developers to implement Change Data Replication using Kafka to provide multi region high availability.
• Implemented grpc services for network management.

Sep 2019 – May 2022 · 2 yrs 8 mos · Pune Area, India

• Achieved tight integration of Presto/Trino with Apache Geode cluster to deliver best performance for OLAP query.
• TPCDS benchmarking.
• Query performance analysis to pinpointed bottlenecks in join operators, scan operators, task de-serialization, metadata initialisation

Apr 2018 – Sep 2019 · 1 yr 5 mos · Pune Area, India

• Designed and implemented SpringBoot REST Application on cloud.(https://misujami.com).
• MiSujami is a joint family tree platform for person to create family mesh (Genealogy).
• Technical stack selection
• Innovated Intelligent numbering system for all close relations considering multiple marriages, multiple relations between 2 close relatives
• Merging of graphs to build a network of all relatives
• Design database considering graph merging/traversing
• Built a team of smart developers and delivered MVP within 8 months.

Feb 2016 – Mar 2018 · 2 yrs 1 mo · Pune

• SnappyData is a spark database where Apache Geode is tightly integrated with Apache Spark.
• Written a paper for VLDB, SIGMOD conferences.
• Tune SnappyData for best performance for large volume of data on large cluster.
• Designed test framework to automate TPCH & TPC-DS benchmarks from scratch
• Tuning overall memory of the product.
• Worked on multiple GC algorithms like G1 ,CMS for different workload using GC log analyser tools, JProfiler.
• Responsible to check and intimate technical gaps between other HTAP products Memsql, SparkSql, Impla/Kudu.
• Tune SnappyData for best performance for large volume of data on large cluster.

Apr 2013 – Mar 2018 · 4 yrs 11 mos · Pune

• Developed following Geode Features.
• Developed AsyncEventQueue to stream events using sliding window protocol
• Developed Multisite WAN replication framework : auto-discovery of site, sending data serially as well as parallely.
• Partition to Node Affinity : feature that provides standard partitioning along with specific data's affinity with specific member.

Jun 2010 – Mar 2018 · 7 yrs 9 mos · Pune Area, India

• GemStone Systems got acquired by VMWARE in June 2010. I continued working with Gemstone Division of VMWARE under R&D group. Working on new features of GemFire.
• Function Execution Service on p2p as well as client server.
• Implemented Stats collection
• Partition Region Rebalancing

Apr 2010 – Present · 16 yrs

• 1. SnappyData : https://github.com/TIBCOSoftware/snappydata
• TPCH/TPCDS Benchmarking Implemenration :
• https://github.com/search?q=repo%3ATIBCOSoftware%2Fsnappydata+kishor&type=code
• SnappyData is a tight integration of Apache Spark with Apache Geode (connector mode is also there).
• I spearheaded the implementation of the TPCH/TPCDS benchmarks from inception, developing a comprehensive framework that allowed for seamless modularity. This framework enabled the benchmarks to be easily rerun facilitating the observation and analysis of performance issues with precision and efficiency.
• 2. Apache Geode: https://github.com/apache/geode
• WAN Replication and Queue Implementation:
• https://github.com/apache/geode/search?q=AsyncEventQueue
• https://github.com/apache/geode/search?q=GatewaySender&type=
• Design, coding, testing. GatewaySenders are used to replicate data from one site to another separated by geographical location parallely (ParallelGatwaySender) and serially (SerialGatewaySender)
• AsyncEventQueue is a Kafka like mechanism for event stream handling.
• Wan auto-discovery :
• https://github.com/apache/geode/search?p=2&q=Discovery
• Discovery of multiple wan site.
• Design, coding, tests, data structures choices are driven by me. Java concurrency utilities are used extensively.
• Fixed Partitioned Region:
• https://github.com/apache/geode/search?q=Fixed
• To maintain the data's affinity to a particular node. Critical aspects of this feature like bucket calculations, bucket rebalancing are designed, coded, unit tested by me.
• PR Single Hop:
• https://github.com/apache/geode/search?q=*hop
• No need for extra hoping with the server cluster to find out the required data
• Partition Region Rebalancing:
• https://github.com/apache/geode/search?q=*rebalance
• Whenever node goes gown OR new nodes joins the cluster, Rebalancing op is triggered through a various mechanism which rebalances data buckets (colocated) in cluster for optimized use

Nov 2007 – Mar 2018 · 10 yrs 4 mos

• I have joined Gemstone System in Nov 2007.
• GemStone Systems is the leading enterprise software company exclusively focused on in-memory data management and distribution. With our patented technology, enterprises can gain ultra fast and reliable access to operational data while dramatically enhancing data scalability and resiliency.
• I am mainly focused on client-server and p2p architecture of partitioned data with high availability of data.

Jul 2005 – Nov 2007 · 2 yrs 4 mos

• I joined Perot in July 2005 as a fresher.
• I got trained on Java/J2EE technology and also completed my SCJP.
• I worked in Travel Vertical where Lufthansa was our client. I was involved in developing Shipment Database and Freight Way Bill for Lufthansa Cargo. I Worked with Lufthansa team in Frankfurt to integrate our modules with their existing modules. Quite challeging job and very nice experience!!!!