A netrii venture · forming

When the physics must be right

Deep-Tech Advisory works at the physics end of computing: data storage architecture, physics-inspired digital twins, and quantum readiness — for decisions where a wrong assumption costs millions.

Led by Sharat Batra, PhD — 50+ patents, 40+ refereed publications, and four decades of taking hardware from first principles to high-volume manufacturing.

Three practice areas

Each one sits where first-principles physics meets a consequential business decision.

Data storage & AI infrastructure

Storage is the quiet constraint under every AI roadmap. We advise on storage architecture and economics with the authority of someone who shaped the industry roadmaps — SMR, MAMR, and HAMR — that today's AI data infrastructure runs on.

Physics-inspired digital twins

Physics-informed neural networks (PINNs) embed governing equations directly into machine learning — so models extrapolate correctly even when data is sparse, noisy, or expensive. Proven in manufacturing: ML-driven test optimization that saved $30M+ in NAND production.

Quantum readiness

Honest benchmarking of what quantum computing can and cannot do for your problem — from an advisor who does the math again rather than repeating vendor claims. Know when to invest, when to wait, and what to prepare.

The AI got the equations right — and the assumptions wrong

That single failure mode, caught in our own AI-assisted physics work, is why deep-tech decisions need practitioners who verify from first principles. We use AI aggressively, and we check its physics ruthlessly.

Fast and reliable, with the right bench

One senior practitioner leads every engagement, and the lab brings in affiliated specialists — practicing physicists, materials scientists, systems architects — exactly as the problem demands.

Judge the work, not the pitch

The practice publishes its methods in the open. Start here.

PINNs: Predicting Projectile Trajectories from Sparse, Noisy Data

A technical brief showing a physics-informed neural network predicting a full trajectory from 8 noisy measurements — where a conventional NN fails completely.

Read the brief

When Physics Rescues Machine Learning

The dispatch behind the brief: why embedding governing equations into the loss function beats pure data-driven learning in physics-constrained domains.

Read the dispatch

The Engineer Who Did the Math Again

An interactive story-course that rebuts quantum computing hype the practitioner's way — by redoing the math, line by line.

Take the course