It’s not hard to see how we added value to prior projects.
Technology: Apache Spark, Scikit
- Transactions Classification - applied machine learning to banking transactions descriptions in order to extract transaction category. Leveraged regulatory data sets to identify sender, recipient and beneficiary of transactions.
- Email Analytics - analyzed email traffic (email size, recipients) and recommended measures to address data breaches resulting from employees sending “print screens”. Established attachment size threshold.
- Anomaly Detection - leveraged unsupervised machine learning models (e.g. clustering) to flag outliers as a proxy for access anomaly and established a baseline for app permission levels.
Technology: Python, Tableau
- Trade Processing - analyzed and identified drivers for straight-through processing (capture, process, repair) rates and recommended opportunities for improvement (e.g. broker data, asset data, internal pipelines).
- Travel Spend - delivered exploratory analysis with the following scope: reduce unnecessary travel, improve advanced booking, increase adoption of online tools, increase economy class incentive rate, increase hotel attachment rate.
- Transfer Agency - evaluated the feasibility of developing products that can be marketed to TA clients. Analysis focused on: investor profile (e.g. sources of health), investor behavior (e.g. transactions).
Technology: Python, Tableau
- Assets under Management Flows - as a result of a Securities and Exchange Commission request, modeled annual change in AUM between flows and market movement. Calculated currency attribution to better understand market performance.
- Prospect Fee Benchmarking - part of the bidding process leveraged public data sets (e.g. US Department of Labor 5500) and benchmarked the fees that a prospect is paying. Produced year over year market analysis.
- At Risk Clients - developed model/prototype to identify at risk clients based on relationship profile (age, number of products), transaction patterns and interactions (e.g. number of emails).
Technology: TensorFlow, AWS, Tableau
- Sentiment Analysis - extracted client sentiment at the document and sentence level using open source deep neural networks models.
- Topic Modeling - identified patterns/topics present in text documents (e.g. client conversations).
Technology: Apache Spark, Hadoop
- Recommendation Engine - prototyped model to recommend products to client facing employees for cross selling opportunities
- Real Estate Utilization - leveraged turnstiles and computer logs to calculate building and floorlevel utilization rates.
Technology: AWS, MongoDB, Spark
- Bitcoin Analytics - extracted and processed all bitcoin transactions to identify patterns. Predicted prices based on order flow.