About me

I train neural networks to make robots autonomous.

[Aug 2025 - Present ] I am the Senior Director of Autonomy at Serve Robotics, where I lead the machine learning team building the intelligence that enables our robots to navigate safely and autonomously through complex, unstructured urban environments. Every day, our robots must reason about traversability across diverse terrain, adapt to changing weather and lighting conditions, navigate crowded sidewalks and social interactions with pedestrians, and obey traffic rules at intersections, all while behaving safely, predictably, and courteously around vulnerable sidewalk users, including people with disabilities.

[Feb 2022 - Aug 2025 ] I was the CTO and Co-Founder at Vayu Robotics. We built autonomous robots that operate in bike lanes and road margins. We built an end-to-end autonomy stack that goes from multi-camera inputs to robot trajectory outputs. We developed a training strategy that made it possible for our model to learn entirely in simulation and generalize to the real world. We deployed up to 10 robots in one Bay Area city. In Aug 2025, Vayu was acquired by Serve Robotics.

[2017 - 2022] I was a Staff Research Scientist at Apple, where I worked on machine learning for perception and planning applied to various embodied autonomous systems.

[2015 - 2016] I co-founded Clarevision Research, along with Ruslan Salakhutdinov and Charlie Tang, where we developed perception neural networks for autonomous driving. The company was acquired by Apple.

[2011 - 2016] I was a PhD student in the Machine Learning Group at the University of Toronto, working with Geoffrey Hinton and Ruslan Salakhutdinov. I co-invented dropout and worked on Boltzmann Machines, video representation models, and transfer learning. During this time, I also spent two summers at Google Brain working on speech recognition (2012) and language models (2013).

[2007 - 2011] I obtained my Bachelor’s in Computer Science from IIT Kanpur, India. I also spent a summer at Microsoft Research (2010) working on natural language processing and another at Ecole Centrale, Paris (2009) as an undergraduate research exchange student.

Publications

• Robust Robotic Control from Pixels using Contrastive Recurrent State-Space Models
  paper pdf code
  Nitish Srivastava, Walter Talbott, Martin Bertran Lopez, Shuangfei Zhai, Josh Susskind
  Arxiv preprint 2021.
• Efficient Embedding of Semantic Similarity in Control Policies via Entangled Bisimulation
  paper pdf
  Martin Bertran, Walter Talbott, Nitish Srivastava, Josh Susskind
  Arxiv preprint 2021.
• Unconstrained Scene Generation With Locally Conditioned Radiance Fields
  paper pdf
  Terrance DeVries, Miguel Angel Bautista, Nitish Srivastava, Graham Taylor, Josh Susskind
  ICCV 2021.
• Uncertainty Weighted Actor-Critic for Offline Reinforcement Learning
  arxiv pdf code
  Yue Wu, Shuangfei Zhai, Nitish Srivastava, Joshua Susskind, Jian Zhang, Ruslan Salakhutdinov, Hanlin Goh
  ICML 2021.
• On the Generalization of Learning-Based 3D Reconstruction
  paper pdf
  Miguel Angel Bautista, Walter Talbott, Shuangfei Zhai, Nitish Srivastava, Josh Susskind
  WACV 2021.
• An attention free transformer
  arxiv pdf
  Shuangfei Zhai, Walter Talbott, Nitish Srivastava, Chen Huang, Hanlin Goh, Ruixiang Zhang, Josh Susskind
  Arxiv preprint, 2021.
• Capsules with Inverted Dot-Product Attention Routing
  arxiv pdf code
  Yao-Hung Hubert Tsai, Nitish Srivastava, Hanlin Goh, and Ruslan Salakhutdinov
  International Conference on Learning Representations (ICLR), 2020.
  
• Geometric capsule autoencoders for 3D point clouds
  arxiv pdf
  Nitish Srivastava, Hanlin Goh, Ruslan Salakhutdinov
  Arxiv preprint, 2019.
• Deep Learning Models for Unsupervised and Transfer Learning
  PhD thesis
  Nitish Srivastava, 2016.
• Unsupervised Learning of Video Representations using LSTMs
  arxiv pdf code
  Nitish Srivastava, Elman Mansimov, Ruslan Salakhutdinov
  ICML 2015.
• Learning generative models with visual attention
  arxiv pdf
  Yichuan Tang, Nitish Srivastava, Ruslan Salakhutdinov
  NeurIPS 2014, oral.
• Dropout: A simple way to prevent neural networks from overfitting
  paper pdf code
  Nitish Srivastava, Geoffrey Hinton, Alex Krizhevsky, Ilya Sutskever, Ruslan Salakhutdinov
  Journal of Machine Learning Research, June 2014.
• Multimodal Learning with Deep Boltzmann Machines
  paper JMLR version supplementary material Code and results
  Nitish Srivastava and Ruslan Salakhutdinov
  NeurIPS 2012, oral.
• Discriminative Transfer Learning with Tree-based Priors
  paper
  Nitish Srivastava and Ruslan Salakhutdinov
  NeurIPS 2013.
• Modeling Documents with a Deep Boltzmann Machine
  paper
  Nitish Srivastava, Ruslan Salakhutdinov and Geoffrey Hinton
  Uncertainty in Artificial Intelligence (UAI) 2013, oral.
• Improving neural networks by preventing co-adaptation of feature detectors
  arxiv pdf
  Geoffrey Hinton, Nitish Srivastava, Alex Krizhevsky, Ilya Sutskever, Ruslan Salakhutdinov
  Arxiv preprint, 2012.
• Enriching textbooks through data mining
  paper
  Rakesh Agrawal, Sreenivas Gollapudi, Krishnaram Kenthapadi, Nitish Srivastava, Raja Velu
  ACM Symposium on Computing for Development (ACM DEV 2010).