Healthcare AI developer for imaging pipelines and medical data
Healthcare teams rarely need a flashy demo. They need imaging data that flows reliably, models that are measured honestly, and an engineer who says plainly what a system can and can’t do. That is how I approach healthcare AI, based on the DICOM and computer-vision work I have built.
- DICOM imaging pipelines at Trinzz
- 95%+ facial recognition accuracy
- B.Tech in AI & ML, CGPA 9.42
The real problem is the data, not the model
Much of the effort in medical AI sits before the model. Images have to come out of their source in DICOM format, be converted into something a model can use, and pass through steps that are repeatable and logged. When that layer is fragile, every result downstream is suspect, however good the model looks. I start with the pipeline, the inputs and the failure cases, and only then talk about models and accuracy targets.
DICOM pipelines and facial recognition at Trinzz
As a software development intern at Trinzz from November 2024 to December 2025, I built AI DICOM medical-imaging pipelines. I also built real-time facial recognition with DeepFace and OpenCV that reached 95%+ accuracy on image and video, and cut load times on the company’s React website by about 15%. The work covered both the imaging data and the computer vision on top of it.
- AI pipelines for DICOM medical images
- Facial recognition on image and video with DeepFace and OpenCV
- 95%+ recognition accuracy in real time
Doctor and clinic data extraction
I built a Practo doctor scraper that extracts doctor and clinic data at scale. Structured provider data like this is useful for directories, market research and outreach for healthcare services. I clean the output into consistent tables and deliver it to Google Sheets, a database or a CRM, and I agree with you up front what is collected and how often before anything runs at scale.
What I will and won’t promise
I don’t hold healthcare compliance certifications, and I won’t label a system compliant on your behalf. What I offer is careful engineering: clear data flows, role-based access with JWT, evaluation against data the model has not seen, and honest reporting of where a model is weak. If your project needs a formal compliance review, I work alongside the people responsible for it.
Rates from $20/hour
Hourly for open-ended work, fixed price for defined projects. The first call is free and there's no obligation.
Frequently asked questions
What healthcare AI work have you done as a developer?
At Trinzz I built AI DICOM medical-imaging pipelines and real-time facial recognition with DeepFace and OpenCV at 95%+ accuracy on image and video. Separately, I built a Practo scraper that extracts doctor and clinic data at scale. I keep claims to what I have actually built.
Can you guarantee regulatory compliance?
No. I don’t hold healthcare compliance certifications, so that decision belongs to your compliance or legal team. I build with clear data flows and role-based access, document how data moves through the system, and make changes they ask for.
Can you work on an existing medical imaging codebase?
Yes. Much of my work is dropping into inherited codebases with no documentation. I reproduce the problem, isolate it and write up the cause before fixing it, which matters more than usual when the data is medical images.
What does a healthcare AI project cost?
Rates start from $20/hour, with fixed-price quotes for defined scopes such as a DICOM preprocessing pipeline or a data-extraction job. Book a free 30-minute call to talk through the data, the goal and the constraints first.