https://owaisbarkati.com/writing Owais Barkati is an AI systems engineer in Mumbai, India. He builds multi-agent AI systems, retrieval pipelines and the high-throughput data infrastructure that runs them — Apache Kafka, Debezium CDC and ClickHouse. en https://owaisbarkati.com/writing/debezium-clickhouse-cdc https://owaisbarkati.com/writing/debezium-clickhouse-cdc Change data capture delivers at-least-once, and an analytical store counts duplicates rather than correcting them. Here is why the binlog beats a batch extract, where deduplication actually belongs, and the two failure modes — schema evolution and silent snapshot gaps — that break pipelines quietly. Thu, 08 Oct 2026 00:00:00 GMT https://owaisbarkati.com/writing/federated-fine-tuning-flower-qlora https://owaisbarkati.com/writing/federated-fine-tuning-flower-qlora Federated learning moves the model to the data instead of the data to the model. That inversion is only practical for large language models because QLoRA shrinks both what each client must hold in memory and what it must send back. Here is why the two techniques belong together. Thu, 08 Oct 2026 00:00:00 GMT https://owaisbarkati.com/writing/langgraph-supervisor-routing https://owaisbarkati.com/writing/langgraph-supervisor-routing One agent holding every tool degrades as the toolset grows. A supervisor that classifies intent first and dispatches to a specialist keeps each agent's context narrow and each failure attributable — but routing becomes the single point where the system can be wrong beyond recovery. Thu, 08 Oct 2026 00:00:00 GMT