conscioustahoe

Hey, Ashutosh here.

I am a deep learning researcher interested in computer vision and medical AI.

Joined MedARC to work on reconstruction of visual perception from brain activity

Joined Genzel Lab to work on unsupervised analysis of brain signals to understand sleep physiology in mice

I created an unsupervised algorithm that can identify sleep stages in mice by analyzing EEG and EMG signals. This method automates the sleep classification process, removing the need for manual scoring, which saves researchers a lot of time and speeds up sleep research.

Joined Neuromatch as a deep learning teaching assistant

Neuromatch was intense and incredibly rewarding. As a deep learning TA, I guided a pod of 10 students through various deep learning concepts, optimization techniques, and architectures. We also collaborated on a project involving a multi-agent reinforcement learning model that examined predator-prey collaboration behavior in biological communities.

Let's build a neural network

Dive into the inner workings of neural networks by implementing everything by hand. Understand backpropagation and gradient descent algorithms, which are crucial for learning. We'll revisit Calculus 101 to manually calculate gradients or derivatives at each step of the process. As our network scales up, we'll automate the entire learning process. This hands-on approach ensures a deep understanding of how these algorithms operate at a fundamental level.

Our startup got acquired by the world's leading cyber security company

Building the startup was an incredible experience. I worked with a brilliant team, tackling challenging engineering and machine learning problems. The hard work paid off when our startup was acquired by Palo Alto Networks.

Google offered me a job in Dublin

After 5 technical rounds, 1 behavioral round, and 3 team matches over 5 months, I finally got an offer from one of the toughest companies in the world to get into.

Joined Gamma.ai as a founding engineer

We are building deep learning models to detect threats in workplace conversations in real-time.