Anmol Garg

At EXL Analytics, I build the data infrastructure that AI systems run on. I built real-time event collection and ingestion on a highly scalable pipeline — PostgreSQL change data captured via Debezium CDC into Apache Kafka, processed with Spark Structured Streaming and Apache Flink across batch and streaming workloads. I also build the pipelines behind AI model training and autonomous agents: feature engineering, embedding generation and vector retrieval on Milvus, and MCP servers that expose governed datasets and tools for LLM access. My ETL and ELT pipelines run on Airflow over Delta Lake and Databricks.

I came to AI from the data side, and I build like it. At Jio Platforms, I built a Spark and Kafka streaming framework on Hadoop for JioFiber’s MyJio app, processing 1 TB of device data so customers could self-diagnose connectivity, Wi-Fi and set-top box issues, and an end-to-end ETL pipeline for Jio AirFiber that cut outage detection time by 30%. That experience left me with a conviction I bring to every AI system I build: it is only as reliable as the data platform underneath it.

I hold a B.Tech in Electronics Engineering and am a Databricks Certified Data Engineer Professional and a Microsoft-certified Fabric Data Engineer Associate. Outside work, I’m a competitive programmer — All-India Rank 1,122 on LeetCode (top 0.08%) and Expert on Codeforces (peak 1617).