active project

AgScribe

LLM-driven structured information extraction over transcribed interviews, using private model deployment and an explicit target schema.

AgScribe grew out of ongoing LUM AI work in agriculture and qualitative research workflows. Publicly, the useful point is the system pattern: transcribed interviews are processed with privately deployed language models to produce structured outputs against defined schemas, making source material easier to inspect, compare, and reuse without naming the underlying customer engagements.

Role
System designer and applied scientist
Period
2024-2025
Audience
research teams, applied NLP teams, domain experts

Themes

Skills and Tools

FastAPILLMsTransformersS3Schema Design

Links

Back to projects