Surbhi Bakhtiyar

• Developed an Ansible-based Kubernetes Operator to manage the lifecycle of on-prem services, automating tasks such as resource management, scaling, self-healing, backup/restore, and rolling updates
• Collaborated with cross-functional teams and research labs, driving enhancements and integrating new capabilities into the platform.
• Implemented real-time data processing pipelines using Apache Spark Streaming, with Apache Kafka as the event streaming backbone.
• Spearheaded customer onboarding efforts, providing hands-on support and ensuring smooth adoption of services while resolving technical issues to optimize the customer experience.
• Led the effort of integrating advanced monitoring and observability tools such as Instana, LogDNA, and IBM Cloud Activity Tracker. Enhanced system diagnostics, performance monitoring, and issue resolution capabilities.
• Onboarded and provided support to over 20 customers, some managing workloads across more than 500 nodes.

• Led the containerisation of all applications using Docker, ensuring consistency across environments. Managed the full lifecycle of containers, including deployment, scaling, and maintenance, using Kubernetes (K8s) for orchestration..
• Built end-to-end REST APIs using Java to automate the creation of Spark clusters with custom configurations, including resource allocation, monitoring of Spark applications, and data plane management. Enabled seamless integration and control over cluster operations through a robust API framework.
• Designed, developed, tested, and deployed scalable microservices. Ensured the smooth rollout of updates and maintenance through automated CI/CD Jenkins pipelines.
• Improved software efficiency by troubleshooting and resolving coding issues.

Performed performance testing on a Hadoop system with real-time streaming Twitter data, focusing on data collection, processing, and visualization to extract actionable insights. Optimized the machine preparation process by leveraging Docker, reducing Hadoop cluster setup time on IBM Cloud from 35 minutes to 7 minutes, resulting in faster cluster creation for end users.