Feb 2022 – Jan 2023 · 11 mos · Hyderabad, Telangana, India · Hybrid

• I was a member of SDIO group (search and data infrastructure org) at Compass IDC where I was working in the expansion team that took care of building and maintaining data ingestion pipelines to bring new data regularly on the backend platform. We were responsible to develop services for data ingestion alongwith debugging and troubleshooting tools in order to ensure data accuracy and high availability.
• My important contribution here was to build a fail-safe and scalable utility to ingest high volumes of historical data from third party MLS servers to our platform in a reliable manner without disturbing the regular flow of incremental data from the same servers. It involved building a service with APIs to submit new jobs and cancel, revive or fetch the current status of a job. The service took care of scheduling new jobs and maintaining multiple jobs running concurrently. Eventually it enabled us to run multiple jobs concurrently with each job ingesting over a million records in a fail safe manner by simply calling an endpoint.

Dec 2019 – Dec 2021 · 2 yrs

• I was part of the Catalog management team responsible to manage images (essentially large binary files) on the designated clusters (mostly required to create new VMs) and facilitate file management and data transfer across the clusters. It involved building microservices and customer facing APIs and building secure platform utilities for inter VM service calls and assessment of end user APIs.
• My important contribution here was to develop new fail-safe APIs to upload large image files (many times greater than 100GBs) to multiple clusters from an external source in small chunks. It protected the users against http connection drops while trying to upload large files over an unreliable network. It was also meant to help customers using some old browsers in their legacy systems that didn’t support upload of a file above a threshold that was clearly not sufficient for modern use cases. I was also involved in multiple problems of streaming data over network gateways and optimizing performance for the same. Notably I was able to improvise data throughput by four times in one such use case.

Nov 2018 – Oct 2019 · 11 mos · Bengaluru Area, India

• I was working on a project aiming building of high precision road maps in an automated manner from data reported from specific devices developed the company. It involved camera calibration, distributed systems to handle large data and solving classic robotics problem like SLAM (simultaneous localization and mapping) problem to detect and map objects on road with high precision.

Aug 2017 – Nov 2018 · 1 yr 3 mos · Pune Area, India

• I was involved in predictive analysis to identify the nature of data we got on our backup servers. I developed statistical models to simulate the backup load to evalute the performance of our internal systems. I was also involved in experimenting and finding out the bottlenecks for various types of loads in our backup service.

Mar 2015 – Oct 2016 · 1 yr 7 mos · Hyderabad Area, India

• As a member of STCI, I was working on Bing serach by Microsoft. I worked on the development of some backend components for Bing rewards program and extracting the user specific data from queries for business reporting.

Jan 2014 – Feb 2015 · 1 yr 1 mo · Bengaluru Area, India

• SCC was an HLS tool from Synopsys. As a frontend engineer, my task was to handle various optimization problems in the intermediate code. It involved some memory usage optimization, handling and maintaining data dependence analysis solutions and some logging tools.

Aug 2008 – Dec 2013 · 5 yrs 4 mos · Bengaluru Area, India

• I was working on the topological models to build wireless network infrastructure and efficient routing protocols on them. I implemented various heuristics to develop such models on large number of nodes. I was also conducting research on the computation of these models to optimize certain parameters.

Jan 2006 – Jan 2008 · 2 yrs · Pune Area, India

• VCS (Veritas cluster server) used modules called LLT(Low latency transport) and GAB(Global atomic broadcast) for connection management, data transfer on parallel channels in a cluster. I was working on development and maintenance of LLT. Some tasks I delivered in this tenure was enabling IPv6 for LLT over UDP and auto detection and recommendation of connection links within a cluster.