Mar 2020 – Sep 2021 · 1 yr 6 mos · Hamburg, Hamburg

• Jimdo is one of early companies enabling small business building website and providing them a digital platform. I have joined the data team, we use redshift as warehouse, with ETL pipelines mostly written in python. Did some work with anomaly dedution of input raw data and worked on predictive model for forcasting the payment using simple linear regression, involved in exploratory data analysis including deployment of model in production.

Jan 2019 – Feb 2020 · 1 yr 1 mo · Hamburg Area, Germany

• Working on streaming solution on flink for anomaly deduction, the plan is to use simple rules to begin with and use data science later on to deduct anomalies in the transaction data related to the fintech application. The Building new reporting platform at deposit solutions.Spark was used for ETL. Developed micro services using kotlin.

Dec 2017 – Dec 2018 · 1 yr · Hamburg Area, Germany

• Worked along side data science team with some interesting work with traffic curation for eliminating traffic which has less potential revenue.
• I used to work with data team , involved writing workflows using airflow to crunch to data from Kafka using spark jobs. I was involved in moving the message broker from Kafka to kinesis, found a critical bug with kpl library in AWS which was causing the service that hosting the kinesis process cpu hit 100%, worked with the AWS team to get that fixed. The data team workflow processes about billions of ad request messages per day or 12 terabyte of data. Working on streaming processing system with flink to reduce the time from source to sink from 6 hours to 15 minutes.

Aug 2016 – Nov 2017 · 1 yr 3 mos · Bengaluru Area, India

• 1. Solved a connection issue with elastic search where the client connection where getting dropped, tuned the ping timeout parameter which solved the issue.
• 2. Debugged a production issue with elastic search, suggested to eliminate the use of dynamic indices to static index so schema changes do not cause the parse failure of data.
• 3. Suggested couple of platform improvement such as adding a metric collection system for tracking the pipeline jobs and export functionality to get the data out of hive to external storage such as redshift/postgresql, these were taken up in subsequent releases.

Jan 2016 – Jul 2016 · 6 mos · Bangalore

• 1. Worked with apache flink to read the data from the kakfa streams and insert to sql sink as aggregates, which is input to peak price algorithm. Got to work closely with data sciences team which was fun.
• 2. Made the execution of peak pricing algorithm horizontally scalable using the redis queue to distribute the load to nodes in the cluster.

Oct 2014 – Dec 2015 · 1 yr 2 mos · Bangalore

• 1. Fixed a complex multi-threaded issue with java android app that we developed for managing warehouse operations, the toughest part of the issue with gathering logs from the devices, which were later moved to elastic search.
• 2. Implemented command pattern in the backend code for bunch of server rest api which cleaned up the readability of the code.
• 3. Implement a data driven way for invoking the next intend in the android UI which provided a greater flexibility to configure the UI without code changes.

Nov 2012 – Oct 2014 · 1 yr 11 mos · Bangalore

• Made an important architecture contribution to the EM service where I suggested to use middleware technology such as JMS queue as opposed to oracle advance queueing, had to prove by suggestion through performance numbers to get the buy in.

Jan 2010 – Oct 2012 · 2 yrs 9 mos · Bengaluru Area, India

• Debugging a really tough multi-threaded issue with transport software which was used to get data from front end to the back end server.

Oct 2007 – Dec 2009 · 2 yrs 2 mos · Greater Philadelphia Area

• Worked the crystal space 3D engine to display the labels for the wires and bond site. The outlines were created using blender a 3D tool, and the sprite was rendered with varying length, width and angle.

Nov 2004 – Oct 2007 · 2 yrs 11 mos · Greater Philadelphia Area

• 1. Fixed a bottleneck issue in performance data collection module there was an accidental sleep in code by a fellow developer was causing the issue, debugged the issue with timestamps from the logs.
• 2. Helped the customer support engineering to identify the root cause of the problem.of RMI exception NoSuchObjectException was thrown when the RMI gets into bad state and the reason for this was a temporary network outage.

Jan 2004 – Jan 2007 · 3 yrs

• I was consultant with velankani working for true position

Jan 2001 – Oct 2004 · 3 yrs 9 mos · Austin, Texas Area

• There was a call trace module which was written by another team and incorrect data was not flowing into our system from that module, so I debugged and helped the team fix the issue, this was outside my area of responsibility.

May 1999 – Dec 2000 · 1 yr 7 mos

• Debugged an interoperability issue of snmp agent with other subsystem, the other team was claiming things are fine from their end, got the tcpdump of response from the component proved there an issue with response, had fun doing really low level debugging.