At Dune Dragon LLC, I built a multi-agent tutoring platform in Python using LangGraph. Its Supervisor-Worker architecture routes student queries to specialized agents and manages conversation state.
I used ReAct and tool-calling for a Knowledge Retrieval Agent that works alongside Diagnostic and Remediation Agents to provide grounded, personalized study plans. I also designed SQLAlchemy models for agent state, conversation memory, and student progress, and visualized class-wide metrics in an educator dashboard.
At Mindlr AI, I built Python FastAPI microservices for LLM-driven agent workflows and developed REST endpoints and React.js components for the Mindlr Assistant. By indexing frequently searched columns, I sped up dashboard data retrieval by 40%, from about 800ms to under 500ms.
My projects include StartupScope AI, a job discovery platform with a Python pipeline ingesting more than 2,000 weekly listings, and StreamFlow, an Apache Kafka event-driven pipeline that reduced data processing latency by 30%. I also implemented a Java REST API and JavaFX client for Compass 360, raising code coverage from 60% to 95% with JUnit tests.
