Accelerate Data Science, AI and Process Automation With Momentum

Sam Ansari
Geek Culture
Published in
5 min readJun 21, 2021


Momentum is a suite of software platforms that enables data engineers, scientists and analysts to efficiently solve machine learning problems and automate business processes.

Fig 1. Momentum AI home page screen shot

Democratize AI Development

Momentum enables end-to-end enterprise automation without any third-party dependency.

  1. A single platform with no coding, and a UI driven approach, builds complex automation tasks rapidly.
  2. No third-party dependencies saves on license costs and avoids integration complexity.
  3. No specialized skills are needed to work with the platform.
  4. An enterprise scale data science platform to train machine learning, computer vision, AI and NLP models that enables intelligent automation.

Momentum Architecture

Fig 2. Momentum Architecture

To enable end-to-end enterprise AI powered automation, Momentum consists of the following four components:

  1. Connect to perform a high speed Extract-Transform-Load (ETL) at enterprise scale.
  2. Machine Learning and AI to rapidly solve machine learning problems by training and deploying models with UI-driven approach.
  3. Automate allows business process automation using intuitive UI based drag-n-drop tools.
  4. 4. Insight to monitor, track and visualize AI outcomes in the form of graphs, charts and dashboards. It also provides a validation and verification workflow engine to manually correct anomalous outputs from AI models.

What Does Momentum Help Do?

  1. Ingest: From all RDBMS, NoSQL, delimited, text, pdf, image, video, audio, sensor, satellite, Restful, medical image, S3, Google Cloud, Dropbox, and more.​
  2. Transform: Highly scalable, simple, UI based engine to transform, merge, join, blend and filter all your data from multiple sources at scale and speed.​
  3. Data Pipeline: Automate data ingestion, transformation, processing and exchange by building customized pipeline to work with data in realtime, scheduled or batch mode.​
  4. ML & AI: With no-coding AI platform, perform automated feature engineering, train AI models and deploy them production.​
  5. Computer Vision: Train and deploy image and video-based classification, object detection, and facial recognition models. Use pre-trained and customize OCR/ICR models.​
  6. NLP: Use or train models for language modeling, text summarization, POS, NER, sentiment analysis, document similarity and more.​
  7. Automation: Create digital workforce by automating repetitive tasks. Utilize AI models to build complex business process automation using UI driven approach.​
  8. Visualization: Visualize AI outcomes, monitor and track KPIs using intuitive web-based dashboard with graphs and charts. Perform validation and verification with customizable workflow.​

Platform Specification

Data Sources

Momentum supports the following data sources for ETL input & output

  1. RDBMS: MySQL, MSSQL, Oracle, DB2, Postgres and all JDBC enabled RDBMS.
  2. NoSQL: Cassandra, MongoDB, MarkLogic, Solr, Elastic Seacrh, and more.
  3. Structured Files: CSV, TSV, Text, XML and JSON
  4. Unstructured Files: Text, images, videos, audios, sensor and satellite data
  5. Distributed File System: HDFS, Google Cloud Storage, S3, and Dropbox
  6. Pluggable architecture to add more sources.

BI Integration

  1. Momentum Insight
  2. Tableau
  3. Qlik
  4. Power BI
  5. Jasper
  6. Micro Strategy
  7. SpagoBI

Built-in Transformation Functions

Mathematical Functions

  1. round(), floor(),(), ceiling()
  2. rand(), exp(),ln(), log(),log2(), pow()
  3. sqrt(), hex(), unhex(), abs(), pmod()
  4. sin(), asin(), cos(), acos(), tan(), atan()
  5. degrees(), radians()
  6. positive(), negative(), sign()
  7. e(), pi()

Aggregate Functions

  1. count(), sum()
  2. avg(), min(), max(), variance(), var_pop(), var_samp()
  3. stddev_pop(), sdtdev_samp()
  4. cov_pop(), covar_samp(), corr()
  5. percentile(), percentile_approx()
  6. histogram_numeric(), collect_set()

Date Functions

  1. from_unixtime(),unix_timestamp(), to_date()
  2. year(), month(), day(), hour(), minute(), second(), weekofyear()
  3. datediff(), date_add(), date_sub()
  4. from_utc_timestamp(),to_utc_timestamp()

Conditional Functions

  1. if()
  3. CASE .. WHEN .. THEN .. END

Machine Learning Algorithms

Supervised Regression

  1. Generalized Linear Regression
  2. Linear Regression
  3. Random Forest Regression
  4. Decision Tree Regression
  5. Deep Learning/ANN Regression
  6. String to Index Model
  7. Recurrent Neural Network Regression(LSTM)
  8. Gradient-Boosted Tree (GBT) Regression
  9. Survival Regression
  10. Isotonic Regression
  11. Factorization Machines Regression

Supervised Classification

  1. Logistic Regression
  2. Decision Tree Classifier
  3. Random Forest Classifier
  4. Deep Learning/ Artificial Neural Network/ Multilayer Perceptron Classifier
  5. Markov Chain with Neural Network
  6. Convolutional Neural Network (CNN)
  7. Gradient-Boosted Tree (GBT) Classifier
  8. Linear Support Vector Machine (LSVM)
  9. Naive Bayes Classifier
  10. Factorization Machines Classifier

Unsupervised Machine Learning

  1. K-Means Clustering
  2. Latent Dirichlet Allocation (LDA) Clustering
  3. Bisecting K-means Clustering
  4. Gaussian Mixture Model (GMM) Clustering
  5. Power Iteration Clustering (PIC)

Natural Language Processing (NLP)

  1. Word2Vec
  2. Document Similarity
  3. Tokenization, Sentence segmentation, POS, NER and concept categorization
  4. Text Summarization
  5. Sentiment Analysis

Recommender Engine / Collaborative Filtering using Alternating Least Squares

Computer Vision

  1. LSTM for OCR and ICR
  2. Convolutional Neural Network (CNN)
  3. Object Detection Using Single Shot Multibox Detection (SSD)
  4. Object Detection Using YOLO
  5. Object Detection Using RCNN, Fast RCNN, and Faster RCNN
  6. Facial Recognition

Feature Engineering

  1. Pearson’s Chi-squared
  2. Correlation Coefficient — Pearson and Spearman
  3. SMOTE
  4. String to Index
  5. OneHotEncoder
  6. Imputer
  7. PCA

Momentum As A Data Platform

Fig 3. Momentum Data Platform Architecture

Momentum Connect allows to automate data ingestion, transformation and processing by creating data pipeline using intuitive UI. In addition to streamlining data wrangling for machine learning, it also allows to build scalable data lake, that is resilient with built-in support for replication, fault tolerance, failover and high availability.

Advanced Feature

Security: Momentum cluster is deployed within a protected and secured network infrastructure.

Privacy and Access Protection: The data models, insights, and output level protection and access are managed through the role-based and sharing mechanism.

Monitoring and Alerts: Builtin support for data and process auditing for failure/success, system monitoring, notification and alerting.

Optical Character and Handwriting Recognitions (OCR/ICR): Momentum supports training custom models to recognize printed and handwritten texts in virtually all languages.

Feedback for Retraining: Momentum Insight provides a customizable web based verification-and-validation (vnv) engine to pipe the anomalous output for manual correction. The manually corrected outputs may be given as feedback for the model to retrain and improve the accuracy.

Getting Started with Momentum

Momentum is accessible via interactive web-based user interface. To access, signup for a free account by clicking the link:



Sam Ansari
Geek Culture

CEO, author, inventor and thought leader in computer vision, machine learning, and AI. 4 US Patents.