Currently developing the workflow orchestration and tools that enable Linkedin engineers to reliably deploy their software in an automated fashion.

I worked on building TFaaS (Terraform as a Service) and a Terraform provider ecosystem to manage Linkedin's public and private cloud infrastructure at scale.

• Designed and implemented cloud architecture that supports running Stratifyd’s production web application, machine learning micro-services and system monitoring components.

• Cloud Infrastructure

• Designed and templated cloud infrastructure from scratch using Terraform and Packer. Reduced manual effort for cluster preparation time from 7 days to hours.

• Designed and implemented monitoring framework for data analytics system. Framework is used to diagnosed system error, traced service request latency and ML training progress.

• Developed and maintained tools and workflows to automate cluster scaling, server-patching, secrets management.

• Designed and customized on-premise cloud infrastructure for client with different use cases.

• Data Integration and ETL Service

• Developed web service based on Tornado to retrieve and process multi-schema data from multiple sources concurrently, including AWS S3, SFTP and 3rd party APIs.

• Designed and implemented Speech to Text audio data transcription parallel processing pipeline using SQS queue with micro-services of machine learning models.

• DevOps Pipeline

• Redesigned and implemented the CI-CD pipeline responsible for building and delivering all agile development code 24/7 to test and UAT environments. Facilitated 50% in end-to-end product development lifecycle.

• Improved performance of container build process by parallel processing jobs, adding compliance security scanning and runtime-only application certificate retrieval.

• Database Configuration and Migration

• Configured PostgreSQL, MongoDB to support multi-tier environments with high availability. Reduced 30% connection loss and ensured zero downtime. Performed database scale-out from single point server to replica set while ensuring data accessibility and consistency.

• Migrated terabyte size production database cross region from US to Europa for existing customers.

• NLP & Predictive Modeling: Clean, processed text data with Spacy, nltk. Built and tested predictive models with Keras, Tensorflow, and Microsoft Azure ML Studio. Models were used to analyze unstructured text and were put into production to assist fund managers’ decision making.

• Prototyping: Created Minimum Variable Product (MVP) and presented to related parties (portfolio managers, analysts and risk management, external clients), collecting feedback and highlighting unique business opportunities.

• Communication: served as an internal data science champion, gave technical talks to business partners and got executives’ buy-in to develop data science analytics.