@anoojpatel

Hi there!

I'm Anooj, a research engineer currently working at Freenome leading the Deep Learning Engineering efforts. Currently I'm focused on scaling large neural networks and novel genomic epigentic foundation models to one of the world's largest cfDNA cancer and disease dataset.


What I'm Doing Right Now!

Freenome

Senior Machine Learning Research Engineer solving Cancer with ML!


Experience

E Ink Corporation

Consulted on projects related to dynamical modeling of multi-colored electrophoretic ink capsules with deep learning models. Goals included building interpretable variability across production batches while understanding the necessary dynamics to drive display color efficiently.

Insight Softmax

Quantitative Research and Datascience consluting on the side. Past clients:

Beacon Biosignals

Developed deep learning models to solve the biggest problems in neuroscience with the smartest experts on the largest brain datasets!

Applied Technology Solutions

Part-time, building out knowledge-graph based persistence framework and researching interesting problems like Distributed ID-Generation, Automated Precision Healthcare, and some more ML stuff.

IBM Watson

Work as an Intern for 2 summers for the Open Source Data and AI Team. Contributed to Spark (Keras to SystemML model conversion engine), working on Model Asset eXchange, and doing some research on GANs for time series data (audio and financial).

the<WIT>Project

Working with some inspiring people and even more inspiring fellows! Mentored and PM'ed a cohort of young women to build cool software engineering projects with non-profits such as the Urban Institute.

NSV Tech

Was an intern. Built out the premiliary models and developed some cool workflows


Publications

Unlocking deep learning for cell-free DNA-based early colorectal cancer detection

Michael Widrich, Anooj Patel, Peter Ulz, Kaitlyn Coil, Thomas Royce, Jimmy Lin, Richard Bourgon, Anindita Dutta; Clin Cancer Res 1 July 2025; 31 (13_Supplement): A045.


Current Projects

Metaxu Language

Currently building a modern functional-style programming langauge for systems programming. It features algebraic effects, a modal system for borrow checking (like Jane Street's Oxidized Ocaml) and algebraic subtyping via SimpleSub. Currently the first iteration is written in Python and compiles directly to C.


Past Projects

Dark Chess Engine

Built the first dark chess engine utilizing some of the great imperfect-information Nash equilibrium solving algorithms, like Counterfactual Regret Minimization and Multi-Objective Information Set Monte-Carlo Tree Search.


WIP: Blog soon™.