Manuel Valera, Ph.D. Email & Phone Number

Marietta, Ga, Us

San Diego, California, Us

- Optimized auto-decisioning rules for Automobile lending, increasing auto-approvals by 163%, generating additional $6M in loan origination in two months.- Calculated PD curves and successfully modeled Automobile portfolios credit risk rating, to compare and convert different credit scores effectively.- Optimized datasets for third-party machine learning models, improving record volume by 500% and establishing data quality standards.

San Diego, California, Us

- Development, testing, validation, and deployment of Machine Learning models for: Fraud detection and risk management scorecards, time series forecasting of expected loans funded, recommendation engines and credit risk modeling.- Conduct high profile, advanced analysis of financial datasets for executive information and critical decision recommendation.- Scouted, researched and contracted high-technology solution vendors for development of cutting edge models.- Proposed novel workflows for data science project development from conceptualization to deployment and beyond via the use of APIs and CI/CD paradigms.- In-house peer-review validation of the external models contracted by the bank.- Assistance on ERM validation efforts to own and third-party models. Job duties:- Research AI and financial services market conditions and assist in conducting feasibility studies to determine market demand for existing and new AI products or services- Analyze and identify business challenges and opportunities and apply modeling/statistics to make strategic and tactical recommendations - Consult with business partners to develop strategy to address and influence business problems- Analyze data, build BI tools and develop financial models - Communicate results and recommendations to senior management and business partners o Identify opportunities to apply quantitative methods to improve business performance - Communicate technical subject matter clearly and concisely to individuals from various backgrounds

San Diego, Ca, Us

Part of the Coastal Ocean Modeling research group at the Computational Science Department.- Researched, designed, and implemented numerical algorithms to improve the General Curvilinear Coastal Ocean Model (GCCOM).- Parallelized and re-validated the GCCOM model for a 6 million grid points experiment, with up to 320 cores across 12 nodes- Modeled field-scale coastal ocean model experiments.- Developed and Implemented machine learning models for dissolved oxygen estimation on highly nonlinear regimes.- Developed and validated time series forecast of dissolved oxygen using neural networks long-term short memory.Additionally:- Managed the Coastal Ocean Dynamics research group.- Teached and graded COMP521 course during Spring 20-21

Boulder, Colorado, Us

Worked with Peter Lauritzen in the Remapping and Conservation Project.Final presentation can be seen at: https://www.youtube.com/watch?v=s8YVDO7RPNo&feature=youtu.be This talk was presented at the Supercomputing '17 conference.

Hired as part of Tsunami Project to modelate scenarios of risk in the Caribbean. Monitoring and improving the automatic earthquake detection system of FUNVISIS. Improving seismological divulgation resources. Remaking and improving the Seismological Bulletin and Preliminary Reports. Development of Android applications for seismological research and divulgation, as well as early warning of events.Main Research:Comparative evaluation of three Tsunamis models: Goto-Ogawa, GEOCLAW and Delft3D for 1867 Virgin Islands scenario and its affectation in the western venezuelan coast.AbstractTsunami numerical simulation has become a powerful tool for decision making in an event of a Tsunamigenic earthquake. However, for the Caribbean there are no historical sea level records to validate those models. Even so, model errors are currently inevitable, due to uncertainties in the initial, boundary conditions and bathymetry, as well as in the oversimplied or entirely neglected physical pro-cesses. In this sense, and for tsunami simulation particularly, dierent methods exists to solve the equations for shallow water, nevertheless each kind of solu-tion presents numerical errors. In this work, a sensibility analysis is presented for three numerical schemes: Goto-Owaga (supported by UNESCO), Delft3D and GeoClaw; using as study case the 1867 Virgin Islands tsunami, according to the parameters of the UNESCO exercise Caribe Wave 2011, with impact on Venezuelan coasts, specically in the Nueva Esparta State. Synthetic water level time series from several locations around the venezuelan eastern coast are ob-tained from the dierent models and compared with the maximum wave height reported by UNESCO for its validation using the NTHMP tsunami model com-parison criteria. The goal of this work is not to select the best possible model, but to make gain of the best of each one to obtain the best estimation of maximum height and arrival times of the tsunami to venezuelan coasts.

Miranda, Ve

Undergraduate Thesis in fabrication and dielectric characterization of polymer-clay nanocomposites. Habilities in machine operation and analysis of nanocomposites in Transmission Electron Microscopy, X-Ray Diffraction and dielectric technicques TSDC and TSPC. Nanocomposites fabrication by solution.

Doctor Of Philosophy - Phd, Computational Science

Doctor Of Philosophy - Phd, Computational Science

Doctor Of Philosophy - Phd, Computational Science

Licenciatura En Física / Physics Bachelor (5Yr Degree), Física