An AI application that turns raw transcriptions of Canadian Parliament sessions into clean final transcripts with accurate speaker labels.
Raw transcriptions of long parliamentary sessions are noisy, hard to read and inconsistent about who is speaking. Before they can be published or analysed, the text has to be cleaned up and every passage attributed to the right speaker.
Built an application that takes the raw transcription of a session and produces a clean final transcript.
Used Claude and OpenAI models to clean up transcription errors and noise.
Labelled every passage with its speaker, so the transcript can be searched and quoted reliably.
Designed the workflow for full-length sessions, with output structured for publishing and downstream analysis.
Transcript refinement pipeline
Speaker labelling
Streamlit interface
Django backend
Digital health, Medical documentation
The AI pipeline behind a medical-visit recorder: one consultation recording becomes an accurate transcript, a plain-language visit summary, and clinician formats such as SOAP notes and referral letters.
Financial services, Wealth management
A natural-language Q&A interface, built on Claude through Amazon Bedrock, over proprietary investment notebooks and financial content stored in S3.
Product innovation, R&D
Primary AI development partner for Narratize's earliest generative features: document ingestion, technical briefs, research reports and scientific analysis.
We can scope and build a system with comparable capabilities, shaped around your own workflows, data and integrations.
