Priyanshu Khandelwal

• Implemented auto-containerization in the WLCG workload scheduling system for 300K+ cores (handling million jobs/month) reduc- ing high wait time for jobs, solved under/over resource utilization efficiency. • Created a platform from scratch for auto-managing infrastructure. Created microservices in Golang, API-Clients, and tools in Python for Rundeck, Foreman, PuppetDB, Openstack etc • Implemented new features in a distributed web service (and it’s CLI client) holding users to schedulers mapping.

• Worked in collaboration with IBM Research, New York and The Open Mainframe Project. • Enabled IBM Z as a backend for MlModelScope (platform to facilitate evaluation and measurement of ML models within AI pipelines). • Implemented byte swapping approach to allow inferencing of little endian ML models on big endian systems in TensorFlow. • Modified the process of serialization & deserialization of MXNet Gluon models to make it endian independent.

• Worked on Kubernetes to develop external Container Storage Interface (CSI) Drivers for Microsoft Azure Cloud Platform. • Implemented CSI specs, and converted Kubernetes in-tree code to out-tree for Azure Disk CSI Driver in Golang.

Automate CERN Cloud Infrastructure on Oracle Cloud and Openstack

Deep Learning Research Internship • Developed software for generating meaningful summary from undocumented source code (multiple programming lan- guage support). • Developed software for providing recommendations to reuse code snippets from existing codebase for performing desired tasks using Variational Autoencoders and Seq-to-Seq Modelling techniques.

• Developed an Android App for Event organisers. Migrated the architecture from MVP to MVVM, implemented background offline job scheduling, QR Code scanning module etc. Released the app on Google Playstore and FDroid. • Developed endpoints for JSON:API Spec Compliant Open Event REST API Server. (Report)