AI/ML Researcher
Resume:
Amirali Khannejad
********@****.*** j 415-***-**** j LinkedIn: Amir Khannejad j
**** ******* ***, ********, ** 94553
EDUCATION
University of California, San Diego San Diego, CA
Bachelors of Science in Applied Physics with Specialization in Materials Physics August 2017 - June 2020 Relevant Coursework: Statistical Learning(Bayesian Decision Theory, SVMs, kernels), Linear and Nonlinear Optimization, Deep Learning(CNNs, RNNs, LSTMs, Auto-Encoders), Quantum Mechanics and Computing, Condensed Matter Physics, Statistical Physics, Computational and Mathematical Physics, Statistical Data Analysis, Numerical Analysis SKILLS
Languages: Python, C/ C++, MATLAB, Mathematica, RStudio, Qiskit, Linux, Git
Cloud Computing: Amazon Web Services
Computational &Visualization Libraries: TensorFlow, PyTorch, Scipy, Sklearn, Numpy, Pandas, Seaborn, OpenCV
Simulation Tools: Simulink, PSpice
RESEARCH/INTERNSHIP INTERESTS
Quantum Computing and Machine Learning:
Independently working on the applications of Feed-forward Neural Networks in Quantum mechanics and machine learning and their application in predicting the state of Quantum circuit and systems in Python and Qiskit
Deep Learning: Interested in applying the fundamentals of Deep Learning algorithms in health care industry, Computer Vision, Low Power Devices
Interested in working on the theories of Deep Learning and Statistical Learning WORK EXPERIENCE
University of California San Diego San Diego, CA
Physics and AI/Machine Learning Research and Projects Jan 2019 - Present
AI/ML Research Internship: Working with an interdisciplinary group of researchers using state-of-the-art technology to conceptualize and refine machine learning algorithms for analyzing and clinically utilizing CT scans.
Covid-19 Pneumonia: Trained a Dense-UNet to segment pneumonia in CT scans. Analyzed resultant segmentation masks for diagnostic and prognostic interests.
Lung Nodules: Trained a 3D Convolutional Neural Network to classify annotated lung nodules on CT scans as malignant or benign.
(June 2020 - Present)
Quantum Computing and Machine Learning: Working on the applications of Feed-forward Neural Networks in Quantum mechanics and machine learning and their application in predicting the state of Quantum systems in Python and Qiskit (Jan 2020- April 2020)
Theoretical Condensed Matter Research: Design and write algorithms pertaining to the structure of condensed matters using theoretical mathematics methods such as twin prime conjecture + modeling scattering patterns in condensed matter using Machine Learning algorithms (Jan 2019 - Jul 2019)
Analog Circuit Design: Responsible for designing various analog filters and circuits Analysis of circuits response and output using MATLAB (Jan 2019 - Jul 2019)
Wyzant Inc. Walnut Creek, CA
Physics and Math Tutor January 2014 - June 2017
Responsibility: Tutored middle school, high school, and college students for math, AP Calculus, AP Physics, and physics Tutored lower and upper division physics, math, and chemistry classes MACHINE LEARNING & DEEP LEARNING PROJECTS
LSTM Network Trained for Music Generation(Trained on GPU):
Implemented a Long Short Term Memory network that learned to play classical music from midi files in Tensorflow.
Added Temperature to the loss function to enhance the probability distribution of the generated content.
Facial Recognition of over 200 Faces Using Deep CNN(Trained on GPU):
Designed and trained from scratch an 8 layer Deep CNN architecture in Tensorflow with top 5 classification accuracy average of 96.3%
Resolved class imbalance problem in the dataset by using Stratified K-fold and resampling of the training batches, and weighted loss.
Implemented Adam Optimizer and Xavier Weight Initialization for faster convergence.
Employed BatchNormalization and Pooling to speed up training.
Multilayer Perceptron Network:
Implemented a configurable neural net with with backpropagation and option to choose optimizer from scratch.
Performed classification on MNIST dataset and achieved 97% accuracy on test set.
Reduced dimension of the data by employing a custom PCA algorithm.
Levenberg-Marquardt Optimizer for Deep Networks:
Utilized the Levenberg-Marquardt Algorithm(LMA) as optimizer for a multilayer perceptron.
Implemented a custom .step method to update the weights
LMA achieved much faster convergence(8 epochs) when approximated a function in R3 compared to (SGD)(25 epochs).
Least Square Classifier:
Classified MNIST dataset by solving a Least Squares problem.
Algorithm learned correct representation of each class of the dataset.
Pre-processed the data to make the dataset linearly independent by removing 0 valued pixels from each sample. LICENSES AND CERTIFICATES
Neural Networks and Deep Learning:
Issuing Company : deeplearning.ai
Issuing Date : Jan 2020
Credential ID : FY7K53BNY48L
No expiration date
Introduction to TensorFlow for Artificial Intelligence, Machine Learning, and Deep Learning: Issuing Company : deeplearning.ai
Issuing Date : June 2020
Credential ID : SX64BGVN9YYA
No expiration date
Convolutional Neural Networks in TensorFlow:
Issuing Company : deeplearning.ai
Issuing Date : July 2020
Credential ID : VUH9RQVRY84R
No expiration date
Biomedical Research - Basic/Refresher Course:
Issuing Company : UC San Diego
Issuing Date : July 2020
Credential ID : 37659613
Expiration Date : July 2023
Natural Language Processing in TensorFlow:
Issuing Company : deeplearning.ai
Issuing Date : August 2020
Credential ID : 6KYXCRNZCSB4
No expiration date
Sequences, Time Series and Prediction:
Issuing Company : deeplearning.ai
Issuing Date : September 2020
Credential ID : M7VGWEB3LK8J
No expiration date
DeepLearning.AI TensorFlow Developer:
Issuing Company : deeplearning.ai
Issuing Date : September 2020
Credential ID : VL9WYB9VXARY
No expiration date
Contact this candidate