André Macieira Braga Costa
- Quant employment verification: CONFIRMED. André Macieira Braga Costa held an actual internship as Quant Research Intern at Giant Steps Capital (Sao Paulo, Brazil), dated Dec 2020 - Feb 2021, where he 'Developed a machine learning model (GANs) to generate realistic synthetic data for enhanced backtesting of trading strategies' and implemented a classifier algorithm. This is a real role (an internship, not a competition, insight day, camp, scholarship, or incoming offer). Identity is supported: the same profile states a Master's in Mathematics (UFMG) and Duke PhD, consistent with the math-olympiad background. Employer classification: Giant Steps Capital is a quantitative investment manager / hedge fund based in Sao Paulo - its function is quantitative finance, not merely a name containing 'quant'. Evidence is index/LinkedIn-sourced, not independent primary-source confirmation.Feb 2021
- Olympiad record (aggregate/olympiad-index sourced, index-only): represented Brazil at the International Mathematical Olympiad (IMO) 2011; gold medals at OBMEP 2006, 2007, 2008, 2009, 2010, 2011, 2012 and at OBM 2009, 2011, 2016. Identity match to the LinkedIn profile rests on name, nationality (Brazil) and mathematics education; not independently confirmed from IMO primary records in this run.2011
- Index-only (LinkedIn enrichment) role: Senior Software Engineer, AI at Orby AI - A Uniphore Company, current (2025-); previously Machine Learning Engineer at the same company (2024-2025, San Francisco Bay Area), building enterprise document-processing and intelligent-automation systems. Index-only, not independently confirmed.2025
- Index-only role: Research Assistant/PhD Candidate at Duke University (2019-2023), leading a research project in computer simulation of hydraulic fracturing; published in peer-reviewed venues.2023
- Education (index-only): PhD, Computational Mechanics/Scientific Computing, Duke University (2019-2023); Master's, Mathematics, Universidade Federal de Minas Gerais (2016-2019); B.S., Mechanical Engineering, UFMG (2013-2018); visiting student, Mechanical Engineering, Rice University (2015-2016).2023
- Index-only role: Software Engineer II at Cadence Design Systems, Belo Horizonte, Brazil (2023-2024), developing a new back-end for a semiconductor design application.2024
- Index-only role: Software Developer at Lawrence Livermore National Laboratory (2019-2023), implementing a 3D fracture solver using the phase-field method and responsible for its maintenance.2023
- Journal publications associated with the Duke/Livermore computational-mechanics work: arXiv 2304.13197, Int. J. Numer. Anal. Meth. Geomech. (Wiley, doi 10.1002/nag.3612), and a 2023 ScienceDirect article (S0167844223003038). Index-linked to the profile.2023
- Earlier roles (index-only): Research Assistant at Rice University (2016), reduced-order modeling of arched structures; Structural Mechanics Intern at ISQ Brasil (2016-2017), Belo Horizonte, inspection/calculation reports for mining infrastructure.2017
Machine Learning Engineer | AI Engineer Building enterprise document processing and intelligent automation systems using all sorts of AI technologies. Contributing to the developm
- Currently develops a new back-end for a semiconductor design application, which improved the performance of database operations in more than 50%. - Improved error handling and
– Developed a machine learning model (GANs) to generate realistic synthetic data for enhanced backtesting of trading strategies. – Researched and implemented a classifier algorithm
- Leads a research project in computer simulation of hydraulic fracturing. - Developed an innovative approach for computational modeling of fluid-driven fracturing. Published in
- Implemented a 3D fracture solver (using the phase-field method), including the design of tests and validation. Currently responsible for maintenance and bugfixes. - Implemented
- Prepared visual inspection reports of structural assets in the mining cable car system Queiróz-Cuiabá, located in Nova Lima, MG, Brazil. - Wrote calculation reports for bolted j
- Developed a reduced order model to predict loss of stability and simulate post-critical oscillations in arched structures. The ROM achieved a 10-fold speed-up compared to standar
Competition record
Brazil · IMO