Aradhya Chouhan

Portland, OR P: +1-201-***-**** ade0mb@r.postjobfree.com linkedin.com/in/aradhya0510 EXPERIENCE

Launchpad.AI – Portland, OR

Senior Data Scientist Jan 2020 – Present

● Working on automated tagging and multimodal representation for the content science team at Nike.

● Built Instance segmentation, Semantic segmentation, Text detection/recognition and other vision based models for content understanding and tagging.

● Scaled models by dockerizing the complete inference pipeline and hosting it on custom built service at Nike.

● Generated multimodal embedding by combining product Images and description for faster and more accurate product similarity search.

● Built custom color palette and proportion detection model to detect exact Nike color palette used in product and marketing content.

Crunch Apps – Newark, NJ

Deep Learning Engineer Apr 2019 – Dec 2019

● Worked on high quality audio synthesis using attention based encoder decoder architectures with prosody encoder for pitch and delivery control.

● Generated innovative input representations for phoneme duration control.

● Built normalizing flow based neural vocoders to get rich time domain signals from mel spectrograms and achieved high quality synthesis at different sampling rates.

● Built Gan based scene synthesis model conditioned on segmaps achieving semantically stable output of images at several resolutions.

Fellowship.AI– New York, NY

Machine Learning Fellow Jan 2019 – Apr 2019

● Worked on Object detection and attention mechanism architectures loke Faster-RCNN, YOLO, Retina Net, etc to detect brand logos in the images from the wild, achieved SOTA on Logos-32 Plus dataset by applying transfer learning at the base ResNet level and incorporating focal loss. New Jersey Institute of Technology - Newark, NJ

Graduate student researcher Jan 2018 – Dec 2018

● Worked on “Fake news detection” by applying hyperbolic language analysis and detecting sentence structuring across a broad range of news articles.

● Studied nonlinear optimization using coordinate descent methods with random restart.

● Studied Johnson-Lindenstrauss lemma by analysing shifts in classification accuracies of models after projection on Gaussian sampled N dimensional hyperspace.

● Also served as Teaching Assistant for a graduate level Machine Learning course. Helped students understand topics and evaluated exams and assignments Yottaasys - Bangalore, India

Consultant Data Scientist Aug 2016 – May 2017

● Innovative feature engineering work on directionally sensitive circular data by applying von mises distribution and circular variance.

EDUCATION

New Jersey Institute of Technology – Newark, NJ Aug 2017 -- Dec 2018 Master’s, Data Science

NIIT University -- Alwar, India Aug 2012 -- May 2016 Bachelor’s, Computer Science and Engineering

ADDITIONAL

Technical: Strong: Python, PyTorch, Tensorflow, Pandas, NumPy Experienced: C, JAVA, SQL Languages: Fluent in English, Hindi

Contact this candidate