Shiv Bhosale developed an end-to-end BERT inference test suite for AWS Neuron within the aws/aws-k8s-tester repository, integrating Docker, Go, and Python to automate setup, execution, and validation of machine learning inference workloads on Kubernetes. His work enabled reliable integration testing by coordinating Dockerfiles, Go test code, Kubernetes manifests, and Python scripts, improving test coverage and repeatability for ML deployments. In the awslabs/amazon-eks-ami repository, Shiv added runtime iptables and NAT support for Amazon Linux 2023 AMIs on kernel 6.12, configuring kernel modules to streamline firewall and networking functionality, and aligning AMI builds with production EKS networking requirements.