At Swagflo, I led the design and development of an ERP-integrated agentic customer support system using LangGraph and Python. It helps users navigate to specific pages, summarize support tickets, generate shipping reports, and create stores, with AWS Bedrock providing LLM inference.
I also architected and built a distributed web crawler and scraper, pairing a Golang/Gin API with Python workers that process jobs from AWS SQS. The services run independently on EC2 instances.
I led development of an embeddable RAG chatbot using Golang/Gin and AWS Bedrock, with OpenSearch Serverless and S3 Vectors for search. By pre-provisioning resources, I reduced knowledge base creation time from 30 seconds to under 100ms; in my projects, I’ve also built DataForgeML and NeuroStash for feature engineering and self-hosted document search.
