Real-time Telemetry Pipeline.

A unified observability platform designed to replace expensive SaaS solutions. It ingests application logs, system metrics, and traces via Kafka, processes them in Go, and stores them in ClickHouse for sub-second analytical queries.

Handling write amplification during traffic spikes. I implemented an adaptive buffering mechanism in the Go consumers that dynamically adjusts batch sizes and commit intervals based on back-pressure signals from ClickHouse, ensuring system stability during 10x load surges.