Jan 2018 – Oct 2018 · 9 mos · Bengaluru, Karnataka, India · On-site

• Rubrik delivers instant application availability for recovery, search, cloud, and development to orchestrate data for hybrid cloud enterprises.
• As a part of Cloud-on team, I got a chance to work on following projects/tasks:
• Projects:
• Cloud-instantiation
• ● Development of cloud-instantiation, a scalable one stop disaster recovery and data migration solution for enterprises to migrate work load from data center to cloud.
• ● Reduced instantiation time from 6hrs to 20mins and a total throughput of 1000 snapshot-conversions per day for incremental conversion of 2*100GB disk VM.
• Resource group Usage optimization
• ● Improved cloud instantiation on azure to reuse specified existing resources and resource groups instead of spawning new ones.
• ● Thereby eliminated the implicit hard limit of 800 instantiations jobs for given user account.
• Cloud Instance type recommendation
• ● Designed and implemented cassandra tables, schema and tool to recommend aws and azure instance types to initiate cloud instantiation job with.

Jul 2016 – Jan 2018 · 1 yr 6 mos · Bengaluru, Karnataka, India · On-site

• Worked with detecting and analysing spam behaviour and click fraud in an online ad network.
• Projects :
• 1. Probabilistic Visitor Spam Signals
• ● Implemented Covisitation signal to find visitors with varied interests based on ad-keyword category and other dimensions and High frequency signals to suspect high volume clicking visitors
• ● Created a Python Project to calculate thresholds based on the underlying distribution, inter-quartile range for outlier detection
• ● Models equipped to work with daily insertions of over several million data points, and runs daily, weekly and monthly data basis.
• 2. Device Fingerprinting
• ● Improved entropy and stability calculation for each signal used for browser fingerprinting.
• ● Based on it, segregated signal lists used for fingerprint calculation based on device classes, on top of previously used single list for all device classes.
• ● Built device fingerprints based on browser fingerprint, location and evercookie collected from repeated user data.
• ● Similarity matching of device fingerprint was attempted with record linkage algorithms.
• 3. Suspect Redis populate
• ● Entities classification built on consolidation spam behaviour of datacenter and other spam signals.
• ● Marked with severity and reason of suspicion, blacklisted with further manual analysis.
• ● Typetagging entities based on kind of suspicion, with updation of everyday data in redis for analysis.
• 4. Cronjob dependancy pipeline with Luigi
• ● Migrated few cron jobs from Ageda scheduler of Node.js to Luigi in Python.
• ● Improved performance, dependancy resolution and run time of cron jobs by distributing among available cores in the system.
• 5. IP size estimation
• ● MongoDB collection for IP size estimation based on number of device classes behind the IP.
• ● Daily updation of size from recent IP activity.
• ● Moving averages and standard deviation of clicks, customers and other factors.
• Technologies/tools : Node JS, Express, mongoDB, Python, flask, Numpy, Pandas, Scipy, matplotlib

Jan 2016 – Jun 2016 · 5 mos · Bengaluru, Karnataka, India

• Motion Sensor Proximity Logout for iOS -
• ● Added logout functionality to windows system, based on iPhone user motion activity of walking away from the system.
• ● Used BluetoothLE, connection strength and user pedometer and accelerometer data as decision making parameters.
• ● Improved spotlight search functionality of existing app to display relevant information on home page and made push notification accessible with 3D touch on iPhone.

May 2015 – Jul 2015 · 2 mos · Kolkata, West Bengal, India

• ● Learnt and implemented application of lattices and linear algebra in security.
• ● Implemented GSO technique and LLL algorithm.
• ● Studied Boneh-Durfee attack on RSA using Lattices.

May 2014 – Jul 2014 · 2 mos · Goa, India

• Customer Complain Web-Application -
• ● Used open source ware XAMPP(MySql, PHP, Javascript and jquery) to develop a web-application which assigns job to an employee when a complaint is logged by a customer.
• ● Major contribution includes database handling and passing of sql query using PHP. Learnt PHP and Javascript for the same.