Shivansh Narayan Email & Phone Number

Bengaluru, Ka, In

Bengaluru, Karnataka, India

Gurugram, Haryana, India

– Designed the event ingestion and search prediction pipeline with new widget creation for a user person-alised home-page . Estimated impact of 3 mil $ / month– Ingestion pipelines handles 1.4 million events /day. Using a fire-and-forget design with a P99 of5ms (MongoDB,Spring-Boot,Thread-Pools)

Autocomplete Migration -> – Led the technical merger of the auto-complete feature of Belvilla into the Global OYO Stack without downtime while coordinating with the business and product team to tune the algo for desired relevance.– Built framework for type-ahead suggestions for multilingual search (6 languages)– The system manages the indexing, relevant searching and updates of 0.6 million locationsML-Feature Ingestion Pipeline ->– Developed an efficient data pipeline for updating the features to be used by ML models. Integration of new features reduced from 3 days to a couple of hours– The pipeline handles feature updates for models catering for India, Indonesia and Malaysia. Feature updates per day are about 9 million (Elastic-Search, Kafka, Apache-Spark)Ranking MLOps Platform ->– We developed and deployed infra for a one-click data preparation and model training/analysis pipeline. Reducing dev effort from 10 days to 3 days and daily cost from 250$ to 150$ (Spark, Airflow, AWS Sage-Maker, Jenkins).• Other Projects: gRPC POC, Spark on K8 POC, Listing Page Widgets

Global Search and Rank Team

Assessment Feature :1. This feature allowed teachers to schedule and create assessments on the platform.2. There were many pre-defined question types along with an intuitive UI for ease.Exam-Engine :1. This service was built to conduct the scheduled assessments.2. Features include -> session management, Auto Save, Auto Submit, Server Side TimekeepingCorrection-Module :1. Module for the correction and report generation/analysis of Submitted assessments.2. This also included a dashboard for the teachers to see the data and its distribution.The System was designed in a way that Assessment Creation, Submission, Checking, Data Analysis happened in an Asynchronous manner.Tech-Stack: ExpressJs , ReactJS , Redux , RabbitMQ , AWS , Docker , Nginx , MongoDB

In Crio Summer Of Doing, student developers build real products in a work-like environment called Micro-Experience, set up by Crio.DoI have built a distributed system backend for food order platform, QEats, with Java Spring Boot and MongoDB.My Project and Skill summary can be viewed from the link below https://criodo.github.io/csod-2019-narayanshivansh49/#CSOD