Nov 2021 – May 2024 · 2 yrs 6 mos · Bangalore Urban, Karnataka, India

• Managing warehouse operations team for all business lines (Standard Delivery, BBNow). Warehouse operation includes receiving skus from suppliers, put into our inhouse structure of racks and bins, picking of order skus and its optimization for last mile delivery. Execute and manage delivery and with 12 team members
• Created the warehouse and supply chain structure in house from scratch in the ecosystem of microservices
• Revamped the warehouse and inventory receiving, replenishing skus via inbound operations and picking operations via inhouse system, It helps to increase efficiency for BBNow instant delivery as well increase in revenue by removing third party vendor tools and process and optimizing the order per vehicle strategy
• Exposed new inhouse warehouse operations system as SaaS offering, it includes making the infra isolated robust and scalable, microservices multi-tenant. It is accessible at https://www.bbmatrix.ai/
• Scale the backend stack to support 10x increase in business combining BBnow and standard delivery after launching in multiple T4 cities as expansion plan
• Revamped existing frontend and backend app for managing farmers and its crop management to increase transparency and reflecting it in BB-Fresho store
• Improved services availability at peak load by introducing proper alert, helm configs, escalation and oncall process

Sep 2018 – Oct 2021 · 3 yrs 1 mo · Pune, Maharashtra, India · On-site

• Designed a new product named Verso for ZS Associates and helped to execute the development and release cycle by leading a team of 6 people.
• Reference
• https://www.prnewswire.com/news-releases/zs-introduces-the-ai-based-verso-product-family-to-enable-intelligent-commercial-execution-300934927.html
• Created MLOps pipeline for scalability for all proprietary algorithms at Organization level for production .Pipeline includes pyspark, Sagemaker, Distributed Tensorflow, alibi for drift and outlier detection, SHAP for explaining, aws
• Scalable Deep Learning algorithm creation and deployment for engagement of sales rep for multi-country, multi-channel marketing to maximize sales. Scalability is achieved through batch based input sequence processing and map-reduce incorporation in distributed tensorflow on ec2 spot instances. Model training time was reduced from 5hrs to 50min at the same infra cost.
• Multiple versions of the above model for different products need to be supported. It was done migrating from the current strategy of Sagemaker ec2 instances to Spark cluster of EMR and managing resources efficiently through yarn. This also results in 40% reduction in infra cost
• Enhanced the genetic algorithm execution time for predicting the sequence of engagement of marketing channels to run in a scalable manner. Improvement of 110x by increasing cost of 6x and making it horizontally scalable. Code level changes include migration from plain python genetic algorithm to pySpark UDF to run in parallel and efficient way
• Multiple POCs related to MLFlow, Metaflow, model training on GPUs, low latency inferencing of ML models through gRPC and nvidia Triton Server were done. Troubleshoot and tune the spark jobs memory issue
• Developed multiple backend APIs over lambda including internal chatbot on dialogflow
• Created a recommender system for recommending insights to end users on mobile App
• Designed a custom framework for deploying API gateway and lambda using java plugin.

Oct 2016 – Aug 2018 · 1 yr 10 mos · Noida, Uttar Pradesh, India · On-site

• Created 3 new modules and redesign some existing modules for Analytics Product Cognowise.
• Created data pipeline for product to ingest data from Learning Record Store (LRS) to process and store in real time using Apache Storm and its middleware using spring for analytics system of student retention.
• Created predictive analytics as a service (SaaS) having machine learning based models in spring with apache spark, mllib
• Created scalable and distributed Alert and Notification service with custom processing on top of Apache Storm.
• Created Data Migration Platform supporting multiple tenants to onboard data from clients legacy system into inhouse data platform to mongodb and HBase using spark and sqoop.
• Harden the server security and Integrated owasp for security on the server side.
• Modification in Caliper/xAPI call pipeline module written in Apache Storm.
• Created an in-house module for monitoring and logging for other teams. ELK stack to dump server logs in elasticsearch.
• Optimize the CI-CD steps and reduce build push time by replacing scp over facebook's Warp speed Data Transfer (wdt) for binary uploads.

Apr 2014 – Sep 2016 · 2 yrs 5 mos · Noida, Uttar Pradesh, India · On-site

• Development of access control(ACL) on the health exchange agent portal web pages for brokers of different roles
• Modification in existing batch jobs for renewal of health insurance of government exchanges ( Obamacare) for insurance renewal

Feb 2012 – Mar 2014 · 2 yrs 1 mo · Gurgaon, India · On-site

• Worked on the flight search and booking part of yatra.com website
• A complete frontend to backend designing and development of online e-ticketing for international flight of live website in PDF format including API Integration as well as exposing it as a web service for other internal system.
• Reduced the render time for Search Flight Page of live website from 3 second to 1 second.(absence of cache) by changing the xslt search pattern of flight details from Amadeus System.
• A Part of core Back-End Team of Travel and did the API Integration with Amadeus cloud for Special Price handling on festive occasions
• A complete frontend to backend designing and development of near search city options in International Search on live website.

May 2011 – Jul 2011 · 2 mos · Tilak Nagar, New Delhi

• 1. Technology - ANDROID
• Site - 'www.webcreations.in'
• Location Tracing and Direction Navigation System with security features
• Developed a mobile Application software on ANDROID that will trace the Location of the phone via GPS service and send it to a apache web-server through REST services. On the web-server the necessary information other than Location (IMEI number and SIM number) gets saved in My-Sql database. At the front-end using the latitude and longitude, a runtime map is drawn via Google map APIs using a PHP script as a web page. On the mobile phone the user also has a feature of Digital Compass that will give the user the current orientation in the form of direction along with a speedometer that shows the speed with which the cell is moving. The software has a security feature, it saves the basic info like IMEI number and SIM number in its SQLite database and each time the mobile get restarted a start-up service check whether a new SIM number is there or not and if a new SIM id get detected then it send all its private information to a specified mobile number through a SMS