Aayush Agrawal

Senior ML Engineer at Bartleby Technologies Pvt Ltd
  • Claim this Profile
Contact Information
us****@****om
(386) 825-5501
Location
Bengaluru, Karnataka, India, IN
Languages
  • English -
  • Hindi -
Skills

Topline Score

Bio

Generated by
Topline AI

0

/5.0
/ Based on 0 ratings
  • (0)
  • (0)
  • (0)
  • (0)
  • (0)

Filter reviews by:

No reviews to display There are currently no reviews available.

0

/5.0
/ Based on 0 ratings
  • (0)
  • (0)
  • (0)
  • (0)
  • (0)

Filter reviews by:

No reviews to display There are currently no reviews available.
You need to have a working account to view this content. Click here to join now

Credentials

  • Convolutional Neural Networks
    Coursera
    Aug, 2018
    - Sep, 2024
  • Improving Deep Neural Networks: Hyperparameter tuning, Regularization and Optimization
    Coursera
    Jul, 2018
    - Sep, 2024
  • Neural Networks and Deep Learning
    Coursera
    Jan, 2018
    - Sep, 2024
  • Machine Learning
    Coursera
    Feb, 2017
    - Sep, 2024
  • Intro to Python for Data Science
    DataCamp
  • Kaggle R Tutorial on Machine Learning
    DataCamp
  • Mastering Data Analysis in Excel
    Coursera Course Certificates
  • Python Data Structures
    Coursera Course Certificates
  • R Programming
    Coursera Course Certificates
  • Statistics with R
    Coursera Course Certificates
  • The Analytics Edge
    edX

Experience

  • Bartleby Technologies Pvt Ltd
    • IT Services and IT Consulting
    • 300 - 400 Employee
    • Senior ML Engineer
      • Nov 2021 - Present

      Created a spell and grammar correction model from scratch. Trained FLAN-T5 fine-tuned language model. The model is trained using 40 million data points. Created docker image and deployment using Kubernetes. Created a traffic model for different sites that classifies whether or not to publish an essay on a website. The model is based on BERT architecture. Developed an API that returns the most relevant image based on the given text. scraped approximately 4 million images and their descriptions from wiki, generated embeddings from the description text using sentence-bert, and clustered them to optimise the search. Show less

  • Pactera EDGE
    • United States
    • IT Services and IT Consulting
    • 100 - 200 Employee
    • Senior AI Engineer
      • Jul 2021 - Nov 2021

      Build shrinkage detection for an American Retail company to identify theft.Used Yolo v5 for product detection and SORT algorithm for tracking theproducts and getting total count in the entire transaction.

    • AI Engineer
      • Mar 2020 - Jul 2021

      Trade Promotion solution for an American food company to predict andoptimize the sales based on promotions for Spain and the Netherlandsmarket. The core solution uses the Autoregressive Distributed lag model(ADL) to predict the sales.Used Genetic Algorithm for optimizing the promotions. The Uplift Sales issplit into the following components: Cannibalisation, Competitor Switching,Category Expansion, Retailer Switching and Category Expansion.

  • nova IQ
    • AI Engineer
      • Nov 2018 - Feb 2020

      Completed a project on the LineHaul Optimisation problem using Deep Deterministic Policy Gradient (DDPG) which is state of the art Deep Reinforcement learning method. Also build a model for speech to text for a client working in defence domain. Completed a project on the LineHaul Optimisation problem using Deep Deterministic Policy Gradient (DDPG) which is state of the art Deep Reinforcement learning method. Also build a model for speech to text for a client working in defence domain.

  • Quantiphi, Inc.
    • United States
    • Financial Services
    • 1 - 100 Employee
    • Machine Learning Engineer
      • Jun 2017 - Nov 2018

      Worked on development of product AthenasOwl. It's a media Content meta tagging product using Deep learning technologies. Have worked on the following projects/tasks: 1. Automated Face recognition pipeline. Use dlib and Inception-ResNet Model for face recognition and classification. 2. Object detection using Retina Net model and achieved mAP score of 40.8 on COCO dataset. 3. Feature Extraction and Image Classification using state of the art deep learning models. Worked on development of product AthenasOwl. It's a media Content meta tagging product using Deep learning technologies. Have worked on the following projects/tasks: 1. Automated Face recognition pipeline. Use dlib and Inception-ResNet Model for face recognition and classification. 2. Object detection using Retina Net model and achieved mAP score of 40.8 on COCO dataset. 3. Feature Extraction and Image Classification using state of the art deep learning models.

  • Leapcheck
    • India
    • Retail Office Equipment
    • Data Science Intern
      • May 2017 - Jun 2017

      Worked in the project of Credit Card Fraud Detection. By using ROSE package produced synthetic data to balance highly skewed credit card dataset containing fraud count of (0.172%) of all transaction. Developed the predictive models using Logistic Regression, trees, random forest and XgBoost. By observing the evaluation metric compared accuracy, precision, recall, sensitivity, specificity, AUC = 0.999 of the model. Worked in the project of Credit Card Fraud Detection. By using ROSE package produced synthetic data to balance highly skewed credit card dataset containing fraud count of (0.172%) of all transaction. Developed the predictive models using Logistic Regression, trees, random forest and XgBoost. By observing the evaluation metric compared accuracy, precision, recall, sensitivity, specificity, AUC = 0.999 of the model.

Education

  • National Institute of Technology Raipur
    Bachelor of Technology (B.Tech.), Mining Engineering
    2013 - 2017
  • Holy Cross Senior Secondary School, Raipur
    2000 - 2013

Community

You need to have a working account to view this content. Click here to join now