Gradient Group’s AI/ML portfolio consists of proprietary work for enterprise clients involving AI/ML strategy, in-house AI talent development, and technical ML implementations in AWS. Check out our enterprise case studies.

AWS Serverless ML Training Pipeline

  • Fully serverless microservices architecture
  • AWS Well-Architected framework reviewed
  • PySpark dataset extraction from delta lake through Glue
  • Data validation, feature engineering & preprocessing, model training, model evaluation, model explainability, and model serving components
  • CI/CD pipeline through SageMaker Studio, CodeCommit, CodePipeline, CodeBuild, and CloudFormation
  • Daily model monitoring Step Function and purposeful re-training
  • Fault tolerance

Custom ML Inference Pipeline (Job Recommendations Engine)

  • Near real-time job recommendations engine
  • Improved user experience resulting in increased revenue
  • Real-time two-way text communication with end users
  • Fully automated solution (no human-in-the-loop or decisions based on analytics dashboard; direct-to-end-user solution)
  • Same CI/CD workflow as the training pipeline

Sentiment Analysis

  • Web Application That Classifies Movie Reviews As Positive Or Negative (Real-Time Inference)
  • End-To-End Deep Learning Pipeline With Amazon SageMaker
  • Custom PyTorch For Model Building, Training, Evaluation, And Modularization
  • Model Containerized And Deployed As A Microservice
  • AWS Lambda, API Gateway, S3, EC2, ECR, ECS
  • 85%+ Accuracy On The Test Dataset For Base Model (25,000 movie reviews)
  • Thorough Unit Testing, Including Neural Network Layer Inputs/Outputs
  • GitHub: https://github.com/LeanManager/SageMaker_Sentiment_Analysis

Image Captioning

Machine Translation

Speech Recognition

  • Deep Neural Network for Automatic Speech Recognition using TensorFlow and Keras
  • Acoustic Feature Extraction (Spectrograms and MFCCs)
  • Testing, Tuning, and Comparison of Various Model Architectures
  • RNNs, GRUs, Bidirectional Layers, CNNs, GPU Support
  • GitHub: https://github.com/LeanManager/Speech_Recognition

Customer Churn Prediction

  • End-To-End Machine Learning Pipeline With Amazon SageMaker
  • Customer Churn Prediction For Mobile Carrier Company
  • Prepare, Clean, and Transform Dataset For Model Training
  • XGBoost Machine Learning Algorithm For Supervised Classification
  • Automatic Hyperparameter Optimization
  • Multi-GPU Cloud Training and S3 Storage
  • Amazon SageMaker NEO Model Optimization

Population Segmentation

  • End-To-End Machine Learning Pipeline With Amazon SageMaker
  • Population Segmentation For U.S. Census Data Using Unsupervised Learning
  • Prepare, Clean, and Transform Dataset For Model Training
  • Principal Component Analysis (PCA) For Dimensionality Reduction
  • K-Means Clustering To Find The Natural And Meaningful Groupings In The Data
  • Analyze And Interpret Model Results To Inform Business Decisions

Restaurant Chatbot

  • Conversational AI Chatbot using Google’s Dialogflow
  • Automated Restaurant Order Placement
  • Dozens of Custom Intents, Entities, and Training examples
  • Speech and Text Dual Support Through Telephony and Web UI
  • Google Cloud Platform Datastore Integration

Human Resources Chatbot

  • Conversational AI Chatbot using Google’s Dialogflow
  • Text and voice-activated chatbot used for getting HR information on-demand
  • Natural Language Processing To Organize Unstructured Text
  • Dozens of Entities Extracted from the Company’s HR Manual
  • Webhook Fulfillment and Omni-Channel Integrations

AI Robot Tracking and Localization

Image Classification

Facial Recognition

Deep Learning Dog Breed Classifier​

Machine Learning Email Classification


  • Proprietary Enterprise Client Work in Deep Learning

%d bloggers like this: