At Hewlett Packard Enterprise, I built long-term memory and skill validation for a ReAct agent architecture, enabling persistent context across sessions and pre-execution checks for selected tools and skills.
I replaced a legacy S3 batch ingestion pipeline with Kafka streaming, sustaining 50K records/sec across Protobuf, Avro, and JSON payloads. I also migrated ingestion pipelines to Delta Lake and designed an AWS Bedrock and LangChain audit pipeline for infrastructure and network logs.
I built an agentic CVE remediation pipeline that connects Trivy scans with LLM-generated patches and automated pull requests. As a Software Development Engineer Intern at Hewlett Packard Enterprise, I built AWS Glue and Lambda ingestion pipelines with PySpark data-quality validation.
In my projects, I built an operational intelligence engine for evidence-grounded incident analysis and QueryMind-DW, an analytical platform with an LLM-powered NL-to-SQL layer. I’ve also contributed to LlamaIndex, dbt-core, and FutureAGI, including work on semantic chunking, validation, and evaluation infrastructure.
