Felipe Giglio
FounderPontificial Catholic University of Rio de Janeiro (PUC-Rio)
Pure Mathematics student, awarded a full scholarship. Motivated by the pursuit of academic excellence and mathematical research. IMO medalist, two-time IMC gold medalist, and recipient of several other international mathematical awards. Author of problems for the Brazilian National Olympiad. Active mentor for Brazilian olympiad students and instructor for a social impact scholarship program.
Over 3000 hours of hands-on experience training the leading frontier AI models, including extensive work with Mercor Intelligence. Deeply familiar with the entire dataset generation pipeline, serving as a writer, reviewer, and math lead while overseeing quality control and originality checks. Specialized in developing rubrics, creating advanced mathematical problems and designing realistic prompts.
João Azeredo
Co-FounderUniversity of Campinas (UNICAMP)
Computer Engineering student at UNICAMP, and FAPESP research scholar in Abstract Algebra and Galois Theory. Multimedalist in major international and national mathematics competitions, including the Ibero-American University Mathematics Olympiad and Brazil’s National Math Olympiad (OBM), and gold medalist in the São Paulo Programming Marathon. Ranked among the top students in Brazil in the IMO Team Selection Tests and actively involved in olympiad training at the highest level. Instructor for Brazil’s official international olympiad teams and mentor to students who later qualified for competitions such as the IMO and EGMO. Specialist in mathematical problem creation and grading, with problems used in official competitions and selection tests, including the OBM and EGMO TST.
Over one year of continuous experience training frontier AI models through Mercor, working across roles including writer, reviewer, and super-reviewer. Deep familiarity with the full dataset generation pipeline for mathematical reasoning models, including the design of high-level reasoning benchmarks, olympiad-level problems, evaluation rubrics, and multi-stage review and quality control. Specialized in developing rigorous datasets and evaluation frameworks for training advanced LLMs in mathematics.
Team
About out expert poolOur team is composed of elite olympiad participants, IMO medalists and math researchers with hands-on experience in AI model training workflows. This rare combination of proven mathematical excellence and direct involvement in model training enables us to deliver high-fidelity golden datasets and expert feedback tailored to the demands of advanced reasoning systems.