Personal.ai

Built personalized LLM agents trained on user-specific documents, chat history, and task logs using LangChain, RAG, and GPT-4.
Developed agent-to-agent communication logic, persistent memory layers using Weaviate and Redis, and
document loaders from internal CMS systems.
Delivered secure multi-tenant architecture using FastAPI, OAuth2, and JWT, enabling private and public agent deployments for enterprise clients.
Improved agent latency by optimizing context windows, async I/O operations, and caching at the vector query
layer.