Autonomous LLM agent integrated into TITAN, a semantic Big Data software platform. It manages the lifecycle of scientific workflows, seamlessly translating natural language requests into complex orchestrated pipelines.
Modern Big Data applications require tools to not only process data but also exploit underlying knowledge. The challenge was bridging the gap between natural language and the complex, semantic composition of scientific workflows (e.g., human activity recognition, land monitoring).
Transformed a basic chatbot into an autonomous reasoning agent. I integrated it into the web platform so it could understand the problem domain and automatically assemble pipelines using TITAN's semantic components (backed by Kafka, Avro, and Spark).
Spearheaded the LLM web integration, enhanced its reasoning prompts, and developed the execution hooks to generate actionable pipeline structures on the TITAN interface.
The agent streamlined workflow composition, proving its validity across academic and real-world Big Data scopes by orchestrating tools without manual node-by-node construction.