I'M

vikramank singh

MACHINE  LEARNING  RESEARCHER  & COMPUTER
SCIENTIST 

I am a graduate student working in the field of Machine Learning and Reinforcement Learning at UC Berkeley, where I am advised by Dr. Ram Akella.  I spent two years at Facebook and at MIT Media Labs working with Dr. Ramesh Raskar on problems in computer vision and deep learning.   

My research interests are in Reinforcement Learning and Statistics. I am most excited about Decision Making under Uncertainty or problems that involve partial observability.

 

Most of my current work is around Bayesian Inference, Hidden Markov Models, Probabilistic Graphical Models, Variational Inference, Particle Filters, POMDPs, Model-based and Model-free learning,  Inverse Reinforcement Learning, etc. 

Relevant Courses I have taken at UC Berkeley:  EECS227A-Optimization Models, CS281A/STAT241A-Statistical Learning Theory, CS294-Machine Learning for Systems, CS294-Deep Reinforcement Learning,  STAT248-Time Series Analysis, INFO271B-Quantitative Research

Other Relevant MOOCs: Stanford CS231n Convolutional Neural Network for Visual Recognition, Stanford CS224N NLP with Deep Learning, Stanford CS288 Probabilistic Graphical Models, Berkeley CS287 Advanced Robotics, MIT18.065 Signal Processing for Machine Learning, MIT6.262 Discrete Stochastic Processes

 

Theory and practice

Technical Interest and Expertise 

Python

95%

++C / C

85%

MATLAB / R

90%

 MongoDB / NoSQL

80%

Git

85%

AWS/ GCP/ Kubernetes

85%

Tensorflow

85%

Keras

95%

Pytorch

80%

Scikit Learn / Numpy / Pandas 

95%

KEYNOTES & TALKS

Projects

My Ongoing Research. See More >
Show More

Experience

2018 July - Present

UC Berkeley

Master's Degree, Focus: Machine Learning

Graduate Student Researcher

Working on various research projects in the domain of Reinforcement Learning and Statistics. 

2019 June  - Present

Ericsson Research

Research Scientist Intern
Working on two research projects - one in the domain of Inverse Reinforcement Learning and another in Secure Federated Learning.

2017-2018

MIT Media Labs

AI Researcher
Trained several pose estimation models like OpenPose and DensePose-RCNN in PyTorch framework to detect yoga poses real-time and deployed it on mobile platforms. Advised by Dr. Ramesh Raskar from MIT Media Lab. 

2016-2017

Facebook, Inc.

Software Engineer (Machine Learning)

Developed an automatic generative algorithm to create street addresses from satellite images by learning and labeling roads and regions. Trained various models like SegNet, U-Net, VGG, and ResNet for segmentation and detection.

2015 - 2016

DRDO, Ministry of Defence, Govt. of India

Undergraduate Researcher

Developed unsupervised learning algorithms to identify important features in order to detect workload in soldiers’ brains using 14, 64, 256 channel EEG data.

2014 - 2015

Indian Academy of Sciences

Research Fellow

Used a core-periphery structure model to detect the spread of virality in an online social network. Developed a Preferential Attachment Model to demonstrate real-time Facebook Graphs and tested those models on a citation and the collaboration network.

2012-2016

University of Mumbai

Bachelor's Degree, Computer Science

Courses Taken: Linear Algebra, Probability and Statistics, Multivariate Calculus, Algorithms and Data Structures, Operations Research, Artificial Intelligence, ERP and Supply Chain Management, Big Data Analytics

CONTACT

I'm graduating in May 2020 and open to full-time opportunities in research. Get in touch if my profile looks like a match! 

vikramank@berkeley.edu 

 
 
 

© 2019 Created by Vikramank Singh