Flexrad ยท 2025 to now

FlexReport

Explains 50 to 200+ medical terms per report.

Role
Founding product engineer
Team
Flexrad
When
December 2025 to now
Stack
Flask, Celery, RabbitMQ, PostgreSQL, Llama 3.3 70B, BioBERT, SciSpacy, LangChain, React, TypeScript

Liver shows mild . A small, likely . Kidneys .

Tap an underlined word.

A sample sentence and a small fixed dictionary, not the real pipeline.

The problem

Medical reports are written for doctors. The people they describe often cannot read them.

What FlexReport does

You upload a report, and FlexReport explains it. It finds 50 to 200+ medical terms in a typical report and gives each one a plain-English explanation.

How it works

Reports go through an asynchronous pipeline built on Flask, Celery, RabbitMQ and PostgreSQL, with Llama 3.3 70B doing the explaining. Text comes out through a fallback chain: PDFPlumber first, then AWS Textract, then Gemini 2.5 Flash, so even scanned PDFs work. Terms are found with a hybrid of BioBERT, SciSpacy and LangChain, each with a confidence score. A vision step also finds where one report ends and the next begins inside a multi-page document.

The React and TypeScript front end shows live progress while a report is processed, takes payments through Razorpay, and exports the result as a PDF.

Result

FlexReport reads scanned PDFs as well as digital ones, and explains 50 to 200+ terms per report.

Next case studyDigiPath Insights