Get to know what we do

ML Engineering

ML Engineering

ML Engineering is gluing together data pipeline, training, prediction and lastly an action. We specialize in productionizing machine learning models in large scale data centers or in edge computing. We have over 10 years of experience in Quantitative Finance specializing in ML Engineering. Data driven decision making is the future of business. Let us build your next project either on a distributed system or on cloud services like AWS, Google Cloud Platform, Azure.

  • Python

    OpenCV

    Scikit-learn

    Jupyter Notebook

    R

    MATLAB

    SQL

    OpenNN

    TensorFlow

    Keras

    PyTorch

    Java

    Google ML Kit

    MLflow

    Amazon SageMaker

    Microsoft Computer Vision AI

    Amazon Rekognition

    Google Cloud Vision API

    Azure Face API

  • AR Models ( ARIMA, ARCH)

    Dynamic Linear Regression

    Neural networks

    Classification models

    Linear regression

    Bayesian model

    Deep learning

    Supervised learning

    Unsupervised learning / Clustering computing

    Decision trees

    Learning to rank

    Logistic regression

  • Data mining: Data collection from various data sources

    Data transformation: Transform data from various sources to build features

    Training and Prediction

    Data comprising: Merge prediction with in house data

    Deliver Data: Save or deliver outputs on disk or deliver to a cloud service

Data Engineering

One of the earlier errors a growing business make is on their data infrastructure. Often there are no central repository of data but compartmentalized and product specific databases. As the business scales up this sort of database solution doesn't scale very well.

We design central data repository. Engineer and transform traditional (SQL, Oracle, etc) database into newer and faster NoSQL(parquet, delta, avro) data store at enterprise level. The solution can be applied both on-premise or on the cloud.

  • Wide data vs long data
    Big data Visualizer/Explorer
    API integration requirements Product(program) dependencies
    Data robustness for computation

  • Apache Parquet, Avro, Delta, Feather

    Apache Cassandra

    Apache Hadoop

    Apache Spark

    Azure data factory

    BigQuery

    Java

    MATLAB

    Python

    R

    Snowflake

    Databricks,

    Snowflake

    AWS redshift

    NoSQL

  • On-prem vs cloud strategy

    Automations

    Big data

    Data acquisition and onboarding

    Data extraction, transformation and load (ETL)

    Data flow

    Data manipulation

    Normalization

    SQL, ORACLE, PostgreSQL -> NoSQL

Time Series Model

We can scale up or take on your next enterprise level time series model. Our in house quantitative researcher/engineer has almost a decade of experience in finance and time series model development. We can deploy advanced dynamic learning models over large scale data such as financial market data Through ML engineering we can fully automate data mining, training, prediction and execution.

  • Classical Econometric Model

    Trend and Seasonality Model

    Auto Regressive Model: AR, ARMA

    State Space Model: ARIMA, Kalman Filter

    Heteroskedastic & Structural break: ARCH, GARCH

    Vector Error Correction Model ( VECM)

    Autoregressive Distributed Lag Model (ARDL)

    Polynomial Regression

    Bagging models

    Boosted models

    Decision tree regressor

  • R

    Python + Scikit-learn

    TensorFlow + Keras

    Java + Apache math libraries

    MxNet

    DL4J

    Jupyter Notebook

Classification Model

We can scale up your model across millions of data points or take on your next enterprise level classification model to achieve business objectives and to help you make better decisions. We have 10+ years of experience in classification model development and ML engineering to fully automate data mining, training, prediction and delivery of outputs.

  • Customer behavior prediction

    Document classification

    Spam filtering

    Image classification

    Product categorization

    Image sentiment analysis

    Customer churn prediction

    Customer behavior assessment for promotional offers

    Anomaly detection problems such as fraud detection

    Credit card fraud detection

    Credit-worthiness assessment

    Blocked order release recommendation

    Sentiment analysis

  • Logistic regression

    Decision trees

    Random forest

    XGBoost

    Light GBM

    Voting classifiers

    Bagging and Boosted Techniques

    Artificial neural networks

Web API

The final stage of a data driven business is to serve customers. We build REST APIs with speed and concurrency in mind. Whether you're looking to serve your external clients or internal, let us transform your flat files into a robust web service.

Automation

Automation is at the heart business optimization. Everything from daunting tasks to time consuming processes can be automated to optimize how your business runs. Automation takes the pressure off so that your employee can focus on creativity.