At Miratech, I’m working on an AWS Bedrock AgentCore RAG and agentic AI system for an enterprise healthcare client. I designed the Bedrock Knowledge Base and evaluated vector databases for its retrieval layer.
I built a RAG proof of concept as a seven-step Lambda pipeline, integrating Bedrock Knowledge Base retrieval and AgentCore invocation with Bedrock Guardrails. When OpenSearch Serverless access blocked progress, I moved the vector store to Chroma and shipped a working end-to-end pipeline.
I also built a FinOps agent with the Strands Agents SDK that generates AWS cost reports using live Cost Explorer data and CloudTrail lookups. I diagnosed a shared-agent-instance bug that leaked conversation history across requests, and I’m developing an Amazon Connect troubleshooting agent proof of concept.
In personal projects, I built a Strands-based SRE agent that correlates CloudWatch alarms and logs, and a retrieval-routing prototype using FAISS, sentence-transformers, and Ollama. I’m an AWS Certified AI Practitioner, with additional experience in Databricks data pipelines and AWS IVR systems.
