Sai Subhasree Pakina

GCP Serverless

Currently working on a tented project under Technology and Research Org, building highly-scalable and reliable systems in the cloud and infrastructure domains, fault tolerant and resilient stateless and stateful services, reliable message transmission protocols.

Developed CLI and a React web app for Distributed State Service (service for distributed coordination in edge applications) APIs. Implemented gRPC client code for streaming APIs, enabled multiple clients to share a single completion queue for clients scaling and reduced memory footprint, connection overhead by multiplexing streaming requests. Designed and implemented APIs that support dynamic deserialization of complex protobuf messages at runtime in Python and JavaScript, allowing clients to retrieve structured data without storing the descriptor or proto files locally.

• Developed bot to automate business process on Automation Anywhere platform; saved 50 hours of manual work • Created metabot in Python using HTTP library, performed user authentication on SharePoint to retrieve files via web • Scripted business logic in SQL resulting in multi-fold reduction in time from 5 hours to 30 minutes • Worked on three projects from cross functional teams to reduce 75 hours/week manual work which costs 350 dollars/hour; developed Excel VBA programs and Tableau dashboards on Salesforce CRM data

• Brainstormed for a challenging real-world problem statement from the banking domain, elicited scope, requirements from business stakeholders and domain experts; came up with a unique technology-oriented solution • Designed neural network ensemble model to predict loan applicant’s credit risk by incorporating three different neural training algorithms using Keras in Python; attained an accuracy of over 75% • Optimized the model by adding batch normalization layer to balance highly imbalanced data; improved accuracy by 20% • Performed data preprocessing, exploratory data analysis, feature engineering on bank loan dataset; improved efficiency by 15% • Presented recommendation and insights along with market strategy to product stakeholders; bagged second position