Machine Learning

In most cases, data can be used to guide human decisions to great returns and without a large investment.

There are situations where the volume, velocity, and variety of data are such that no person could possibly discern the correct path forward. The value and scale of these decisions is such that optimization can produce massive returns for your business. Those are the cases where machine learning should be pursued.

In those cases, CorrDyn can help your company to:

1. Define the business outcomes you want to achieve
2. Determine whether machine learning is appropriate for your budget, use case, and expected return
3. Develop a timeline, plan, and budget for your machine learning implementation
4. Create a proof of concept to demonstrate the value of a selected system architecture
5. Build a production pipeline to implement machine learning at scale, including data ingestion, data transformation, model prediction, model revision, and pipeline monitoring.

Machine learning encompasses tools that should be utilized selectively. When your company is ready for machine learning, the CorrDyn team can lead you down the path to ROI.

Our machine learning projects produce value.

CorrDyn was instrumental in building out various solutions that reduced back-office operations and allowed our teams to focus effort on more strategic revenue generating activities. We rely on them for their expertise across data and technology, reducing our need to make non-strategic hires to maintain systems. CorrDyn has been critical to our success, and we are extremely impressed by the value and quality of the work we receive.
‍
- CEO, Hedge Fund

How We Measure Success

CorrDyn machine learning engagements focus on:

• Time to Value: How quickly can we demonstrate value from the models we implement?
• Reliability: How can we minimize the time and money your company invests in maintaining the machine learning pipeline?
• Return on Investment: How can we maximize the return to your business from the machine learning pipelines we develop?

We want to prepare you for the next challenge: how to action on machine learning models at scale.

Our Toolkit

We choose the tools that fit the job. We build on:

• Machine Learning Modeling Techniques: XGBoost, Deep Learning, Seq2Seq, Sequence Classification, DeepAR, ARIMA, Random Forests, Random Cut Forests, etc.
• Machine Learning Model Types: Regression, Classification, Recommendation Engines, Search, Time Series Prediction, Anomaly/Fraud Detection
• Machine Learning Pipeline Tools: DataFlow, BigQueryML, SageMaker, Docker, ECS, CloudRun, Lambda, CloudML, Various AutoML Tools, Deployment to Edge
• Clouds: AWS, GCP, and Azure
• Data Pipelines: PubSub + DataFlow, Lambda + Step Functions, Spark, Beam, Hadoop, FiveTran, Stitch, Custom Python
• Storage Technologies:  Relational DBs, NoSQL DBs, ElasticSearch, Block Storage
• Data Warehouses: BigQuery, Snowflake, Redshift
• Business Intelligence Suites: Tableau, Looker, PowerBI, Google Data Studio
• End Results: Front End Application, CRM, ERP, Email, Text, Spreadsheet, IoT System, or anywhere you need your data to land.

Example Projects

Our Machine Learning projects have included:

• Developing an automated research system for a hedge fund
• Building a machine learning model deployment pipeline for a hedge fund
• Developing a model to predict financial transactions for a financial services company
• Building a social media text classification model for a technology startup
• Developing various time series forecasting models for an online education company
• Building demand forecasting models for an e-commerce company