Our client is seeking a team of 4 model developers/technical writers to cover their project scope. Roles will sit in Birmingham, Alabama. Local candidates preferred, or candidates who are willing to travel to work site.
- Two will cover 6-8 Regression models
- One will cover 3 non-regression models
- One person will be leveraged horizontally across the models for QA and to assist with transitioning development from SAS to Python, R, or another language compatible with the client’s big data environment
We are looking for strong model developers and technical writers to supplement our existing team with experience across the entire model life-cycle to assist our client with transitioning models impacted by the cessation of LIBOR. The LIBOR Alternative Project is investigating the impact of replacing the London Interbank Offered Rate (LIBOR) with an industry accepted alternative rate, which are currently under development. The preferred alternative rate is the Secured Overnight Financing Rate (SOFR), with other competing rates also under consideration (such as Ameribor). Certain bank models currently rely on LIBOR as an input and will need to be updated to remove reliance on LIBOR and incorporate a new rate (or rates). There is currently not enough capacity for some of the model teams to complete all the objectives related to model updates under the transition, so these consultants will support the development work required to update the models. The team will work with the model owner to update the models as needed and assist with validation activities as requested. We expect the initial phase of work to run from mid-February until September with a high likelihood of additional work to follow.
Summary of Work to be Completed:
- Scope includes removing the reliance on LIBOR for 6-8 regression models and 3 non-regression models, these are treasury models (portfolio balance and revenue)
- Test replacement variables and transformation from an approved list of available (or propose compelling alternatives for us to consider)
- Remove reliance on LIBOR for model ongoing monitoring or performance testing
- Update documentation to be compliant with internal templates (existing documents will be made available to teams) for any changes
- Update desktop procedures and checklists
- Update the implementation and model output process (including data flows, mapping and output report generation)
- Update any quantitative or qualitative overlay or adjustment logic performed distinct from the ‘pure’ regression model to keep outcomes reasonable from a economic and business standpoint
- Formal documents tend to be about 150-200 pages for regression models and 20-50 pages for non-regression models (though, not all needs to be updated, and is comprised largely of tables and charts)
- Output materials for the regression models are currently generated in SAS and converted to Excel or PowerPoint; however, we would like to consider transitioning our development effort into Python or another language compatible with our new big data environment
- Output materials for the non-regression models are generated in Excel or PowerPoint, though some development work may be done in SAS. However, we would like to consider transitioning our development effort into Python
- Proficiency in SAS, Python or R required. Technical writing skills required.
- Experience with accessing and mapping data within a data lake environment and PySpark preferred.
- Assistance with transitioning development from SAS to Python, R, or another language compatible with our big data environment.
- Prior experience building and documenting and/or validating bank models within SR 11-7 guidelines, experience with a US Regional Bank (SR 15-19 bank is a plus)
- Availability to complete development work and meet agreed upon timelines
- Availability to assist with validation activities by providing documentation and engaging with Regions’ model validation team as needed
- Ability to provide ad hoc support (including providing inputs to validation) post development for 3 months
- Familiarity or rapid ability to learn Cloudera Data Science Workbench and common IDEs for development