NoSQL Big Data Storage Build

I need a production-ready infrastructure that can capture and persist high-volume, real-time streaming data in a NoSQL environment, with MongoDB as the core database engine. The cluster must be architected for horizontal scalability, high availability, and low-latency writes so that, down the line, I can mine the stored data for machine-learning model training without worrying about bottlenecks or data loss. Here’s what I’m expecting: • A detailed architecture plan that explains shard keys, replica-set topology, and how the cluster will expand as data grows. • Deployment of the MongoDB cluster (on-prem, cloud, or hybrid—whatever we decide is best) scripted through infrastructure-as-code so the environment can be recreated in a single command. • A streaming ingestion layer that pushes data into MongoDB in near-real time; if you favour tools like Kafka, Kinesis, or similar queues, spell out how they’ll fit into the flow. • Indexing and data-retention strategies tailored to fast analytics queries while keeping storage costs predictable. • Monitoring, alerting, and automated backup routines built in from day one. • Clear documentation and a hand-off session so my team can operate and extend the system confidently. Acceptance criteria: the solution must ingest a synthetic stream at ≥50k writes/sec with zero data loss, automatic failover must pass a node-kill test, and the full stack must stand up via the provided IaC scripts in under 30 minutes. If you’re fluent in MongoDB internals and comfortable designing for big-data scale, I’d love to see how you’d tackle this.