Now scanning · v.05.04.26

Radiologist
who builds.

Dr Sanyam Katyal. Consultant Radiologist in the NHS, founder of FlowRad AI. I build software that gives radiologists their reading time back — and write about how clinicians can ship.

building  flowrad-ai · h1b-radar · tutor
Dr Sanyam Katyal
FRCR · ROYAL COLLEGE OF RADIOLOGISTS
UHDB
NHS Trust
FRCR
+ EDiR · IRIA
3
Live products
2026
Building in public
Journey

From Manipal to the Midlands.

I trained as a radiologist in India — MBBS and MD Radiodiagnosis at Kasturba Medical College, Manipal. Now I'm a Consultant Radiologist with the NHS at University Hospitals of Derby and Burton, with FRCR and EDiR along the way.

The flight is real, the rest you can see by scrolling.

ManipalIndia · 13°N
DerbyUK · 53°N
Booting globe…
13.35°N · 74.79°E · MANIPAL
52.92°N · 1.48°W · DERBY
Build log

Recent shipping.

A living timeline of writing, commits, and product changes. Auto-updated.

2 days ago
postFRCR Exam Preparation: An Honest Guide for IMGs
A practicing UK consultant's roadmap for IMG success in the reformed FRCR Part 2B exam.
1 wk ago
shipFlowRad AI · v0.7 — voice templates v2
Custom voice macros, faster transcription latency, cleaner template editor.
3 wks ago
postThe Reality of AI in NHS Radiology
A frontline perspective on real successes, vendor failures, and the path forward beyond the hype.
1 mo ago
commitH1B Radar — multi-stage AI risk pipeline
Replaced single-prompt scoring with a structured 4-stage analysis pipeline.
2 mo ago
shipRadiology Tutor — beta launch
AI-powered interactive radiology education platform for FRCR/EDiR trainees.
Projects

Things I actually ship.

Each window below shows a stylised animation of what the product actually does.

flowrad.ai · dictate
FINDINGS CT thorax with contrast. 8mm nodule, RUL apical segment. No pleural effusion. No ▸ Suggest 3-month follow-up.

FlowRad AI

Voice transcription → AI-assisted reporting → custom templates. Watch dictation flow into a structured report.

AI · Radiology · SaaSVisit →
h1bradar.com · risk
Risk score
62/100
Moderate

H1B Radar

Personalised H-1B visa risk assessment from USCIS/DOL data via a multi-stage AI analysis pipeline.

AI · Immigration · SaaSVisit →
tutor.sanyamkatyal.com · mcq
QUESTION 1 OF 25 · CHEST
Solitary pulmonary nodule, RUL apex, smoker. Most likely?
A Pulmonary metastasis
B Bronchogenic carcinoma
C Hamartoma
D Tuberculoma

Radiology Tutor

AI-powered radiology education for FRCR/EDiR trainees. Adaptive MCQs, structured feedback.

EdTech · RadiologyVisit →
An average Tuesday

What I actually do all day.

Radiology, demystified. The four things that fill most reading-room hours, told without the jargon.

Scrolling. So much scrolling.

An average CT chest is ~500 axial slices. We scroll through every one. Yes, with a mouse wheel.

8 mm · RUL

Then — there it is.

Eight millimetres in the right upper lobe. The reason the scan was ordered. Measure, describe, decide.

WW 1500
WL −500
LUNG

Why doctors on TV always look wrong.

Same scan, three reads. Lung window, soft-tissue window, bone window. Different numbers, different worlds.

8 mm nodule

Circle. Caliper. Dictate.

Repeat ~200 times. By 11am you've measured more nodules than coffees you've had. The math doesn't always work out.

Now

What I'm working on this week.

Preparing for the RCR AI Conference (June 2026) — presenting a crossover study comparing FlowRad AI vs Dragon Medical for radiology dictation. Pushing FlowRad toward NHS DTAC compliance. Shipping updates to H1B Radar and exploring radiology education tools. Last updated · April 2026
Contact

Let's talk.

Open to conversations about radiology, AI in healthcare, or building products as a clinician.

sanyam@flowrad.ai LinkedIn