BioStack Raises $9.5M to Supply Data for Training Medical AI

Medical AI is held back not by architectures but by data. BioStack, a startup that calls itself a medical data infrastructure company, announced on October 22 that it has closed a $9.5M seed round. Its goal is to turn hospital data into a system for training and evaluating medical AI models. In a market where clinical LLMs and diagnostic neural networks are multiplying faster than anyone can validate them, that targets a real bottleneck.

A team with academic and cloud backgrounds

BioStack has two founders, and their experience shows how the company approaches the problem. Sanat Mishra has worked at Stanford, Yale, Carnegie Mellon and the Max Planck Institute. These are places where medical AI is studied seriously and where researchers know what real clinical data looks like: incomplete, noisy, written by different people in different formats.

Parth Patwa was previously a generative AI scientist at AWS and is affiliated with MIT. He brings the view from the other side: how to build a scalable pipeline for training and testing models, rather than just publishing papers. For a project at the intersection of hospital systems and ML, the mix of research and engineering experience looks logical.

Why training and evaluation

The company's pitch contains two words that are usually kept apart: training and evaluation. Training is straightforward: for a model to read scans, discharge summaries or clinical notes, it needs labeled examples converted to a uniform format. Evaluation is harder in medicine. General benchmarks such as exam-style tests say little about how a model will behave on the stream of real patients at a specific clinic. Whoever owns both the data and the evaluation method gains leverage over the entire development chain.

That is why the idea of turning hospital data into both "study" and "exam" material looks attractive to investors. If the data layer works as infrastructure, both startups with narrow models and large labs that need clinical validation can build on top of it.

What we know about the round

• Company: BioStack, medical data infrastructure
• Round: seed, $9.5M
• Announcement date: October 22
• Founder: Sanat Mishra, experience at Stanford, Yale, Carnegie Mellon and the Max Planck Institute
• Co-founder: Parth Patwa, former generative AI scientist at AWS, affiliated with MIT
• Product framing: hospital data as the basis for training and evaluating medical AI models

What developers and clinics should watch

The main question for companies like this is how they handle data access and privacy. Hospitals are reluctant to share records, and regulatory requirements for medical information are strict. Whether BioStack's layer becomes a standard or stays a niche service depends on how it builds relationships with clinics and protects patients.

Teams building a medical model should view platforms like this first of all as a source of validation. A run on data close to real clinical practice is more convincing than any public leaderboard. For clinics, such a partner offers a way to bring their archives into AI development under control, rather than handing them over to scattered contractors.

For a seed round in data infrastructure, $9.5M is not a modest sum: it is enough to assemble a team, sign the first contracts with medical institutions and demonstrate a working pipeline. The next milestones are public hospital partners and the first models evaluated on the platform.