Blockchain enables decentralized, tamper-evident, and auditable data sharing among untrusted parties, but deployments face a trade-off between scalability and privacy. Public blockchains expose transactional metadata, risking confidentiality, while privacy-preserving approaches like zero-knowledge proofs (ZKPs) often reduce throughput and increase latency due to proof computation. Scalability methods—Layer-2 rollups, sharding, state channels—typically offer limited privacy, leaving metadata open to inference attacks. We propose a privacy-preserving, scalable blockchain architecture for secure data sharing in distributed environments such as federated clouds, healthcare networks, IoT ecosystems, and inter-bank settlements. The design combines Layer-2 zero-knowledge rollups with a modular Layer-1 settlement layer (Ethereum or Hyperledger Fabric), decentralized storage (IPFS/Filecoin), and fine-grained access control. Transactions are batched off-chain, validated with succinct ZK proofs, and only aggregate proofs and state roots are committed on-chain, enabling confidentiality with high throughput. The architecture, deployed in a Kubernetes-orchestrated environment, supports horizontal scaling, automated failover, and observability via Prometheus, Grafana, and Jaeger. A prototype shows up to 5.8× throughput gains over baseline privacy-preserving blockchains while keeping latency within acceptable limits. Evaluation against three baselines—Layer-1 only, Layer-1 + privacy, and Layer-1 + scalability—using throughput, latency, cost, privacy efficacy, and fault tolerance confirms that integrating privacy-preserving cryptography with scalable rollup architectures is both feasible and effective for secure, high-performance blockchain systems.