Post Job Free

Resume

Sign in

Student AI Researcher

Location:
Los Angeles, CA
Posted:
January 03, 2021

Contact this candidate

Resume:

Puranjay Rajvanshi

Los Angeles, CA ***** adi4no@r.postjobfree.com 510-***-**** www.linkedin.com/in/puranjay-rajvanshi www.github.com/puranjayr96 EDUCATION

University of Southern California, Los Angeles Aug 2019 – May 2021 Master of Science (Hon.), Computer Science, Intelligent Robotics CGPA: 4.0 / 4.0 Coursework: Machine Learning, Robotics, Foundations of AI, Analysis of Algorithms, Information Retrieval and Web Search Engines Vellore Institute of Technology, Tamil Nadu, India July 2014 – May 2018 Bachelor of Technology in Computer Science CGPA: 8.91 / 10 Coursework: Data structures and Algorithms, Discrete Mathematics, Linear Algebra, Soft computing TECHNICAL SKILLS

Languages: Python, C/C++, Java, HTML, CSS, JavaScript Software and Tools: Scikit-Learn, Pytorch, Tensorflow, Keras, OpenCV, Git, numpy, Spark Databases: MongoDB, SQL Server, DynamoDB

PROFESSIONAL EXPERIENCE

Student AI Researcher, Institute for Creative Technologies (University of Southern California), USA July 2020 – Present

• Created explainable AI models for reinforcement learning based models covering both model free and model based

• Successfully created custom decision trees that represented the AI model with a 0% error rate

• Improved the quality of the model by increasing complexity and adding new parameters Student Researcher, Robotic Embedded Systems Laboratory (University of Southern California), USA Oct 2019 – June 2020

• Increased the learning efficiency of drones by integrating reverse curriculum generation alongside Proximal Policy Gradients

• Removed legacy system by shifting the entire codebase from Theano and Rllab to PyTorch and Garage SDE Intern, Defence Research and Development Organisation, India Jan 2019 – July 2019

• Successfully deployed an offline Chinese to English translator, achieved superior tokenization by using SentencePiece

• Leveraged NLP techniques like attention mechanism to build an effective translator SDE Intern, Paytm, India Jun 2017 – July 2017

• Corrected manual third-party sellers’ entries by using machine learning tools such as word2vec

• Programmed REST API concepts which were to be applied in an e-commerce marketplace PROJECTS

SoCal FC (Python)

• Developed FIFA playing bot using behavioral cloning, made the bot play more like a human, OpenCV, Wingui and Tensorflow were used

• Used CNNs for feature map extraction followed by multiple LSTM networks to output the agent’s actions Alpha Go (Python)

• Developed a reinforcement learning based Go playing bot with a team of 6 people

• Designed and parallelized MCTS architecture

Vegetable Disease Detection (Python)

• Trained model on multiple plants - tomatoes and grapes to detect diseases by through image recognition on pictures of crops

• Deployed transfer learning techniques to train the dataset on multiple SOTA models

• Compared these models to get best performance and leveraged data augmentation techniques to increase size of dataset

• Acquired Training and Test accuracy to 99.96% and 98.7% respectively with an F1 score of 0.9876 Real Time Facial Verification and Recognition (Python)

• Applied Multi-Task Cascaded Convolutional Network (MTCNN) for facial detection

• Used triplet loss function and a deep CNN architecture (Inception Model) to train model

• Made model robust to changes - beard and glasses as it evolves with each use

• Introduced capability of face setup in under 5 seconds and integrated capability of handling multiple faces concurrently

• Saved up on space by storing face embeddings in a 128-bit vector Motion Planning with a 6-DOF Manipulator (Python)

• Taught a robotic arm with 6 degrees of freedom to reach and grasp a cup placed on table

• Used Rapidly exploring Random Tree planning to achieve this goal PUBLICATIONS

• Deep Learning based Mobile Application for Plant Disease Diagnosis: A proof of concept with a case study on Tomato Plant. In Applications of Image Processing and Soft Computing Systems in Agriculture (Chapter 10, Feb 2019 issue), IGI Global.

• Deep Learning based Plant Disease Diagnosis for Grape Plant. In Proceedings of Research Frontiers in Precision Agriculture, AFITA/WCCA 2018, pp. 417-420, IIT Bombay, Mumbai, India, October 24-26, 2018.



Contact this candidate