1. Tech lead on GCP Cloud External Manager (https://cloud.google.com/kms/docs/ekm), designed and built the novel key management solutions that anticipated and addressed sovereignty market need. Drove the adoption and built the ecosystem around EKM and interoperable keys.

2. Also tech lead on Gemini chat experience for GCP Security, Privacy and Compliance product usage. Efforts include specialized agents that report, auto-fix security and compliance findings, securing AI traffic.

Tech lead on GCP compliance project Key Access Justifications - from defining product, driving designs to overseeing implementations with solid landings.

Previously worked on several ground-breaking key management service projects including Cloud HSM and Cloud External Key Manager

Worked in a team environment to build prototype of an internal applicant tracking system, utilizing

nodeJS/ExpressJS and SaaS tools, implemented data models using MongoDB and Elastic. Saved operation costby 15%

Architected and deployed the distributed data protection strategy for 700+ virtual machines with different virtual environment consisting of ESXi, vCenter, Cloud Director, docker using Commvault Simpana

Successfully created an internal wiki website using Ruby/Sinatra, JQuery and Bootstrap, used by ~1000 employees

Worked with senior developers on the design of natural language processing and machine learning APIs for 30+ Sina products, including Weibo, iAsk, iAsk Doctor

Developed machine learning systems to classify text into hierarchical labels tested by 1,0000 queries per second

Implemented and evaluated existing algorithms for vector space model, topic model, and deep learning based distributed word representation utilizing nltk, word2vec, crf++ and libsvm

Design new algorithms and systems for information retrieval using lucence and gensim, improved recall rates by 20%

Developed server side architecture that can handle 200,000 pageviews/ day using Python/Django, MongoDB and MySQL

Reduced 30% page load time by implementing caching strategies using Redis/Memcached and task queues using RabbitMQ

Deployed Solr-based internal search engine with 300,000+ indexes and optimized queries by tuning JVM parameters

Reduced cost by 10% by incorporating machine learning components into an original manual-based picture classifying system using scikit-learn