Design and solve challenging math problems that expose weaknesses in large language models.
Create high-quality, step-by-step solutions with clear reasoning and multimodal support (equations, visuals, graphs, simulations).
Collaborate with founders to align problem types with model evaluation goals, particularly in topics where models typically fail (e.g., abstraction, multi-step reasoning, symbolic manipulation).
Help define new evaluation benchmarks inspired by high-school olympiad math, early undergrad curriculum, and theoretical mathematics.
Mentor/coach or review work from other math contributors or junior collaborators.
Requirements:
PhD (or strong MSc) in Mathematics or a closely related field.
Strong problem-solving backgroundcomfortable designing and solving non-standard problems.
Excellent communication and writing skills, especially in explaining step-by-step mathematical reasoning.
Experience with Lean tool (highly preferred) or Coq, Isabelle, HOL Light, or other similar theorem provers
Familiarity with LaTeX and math visualization tools (Desmos, GeoGebra, Python/Matplotlib, etc.).
Nice-to-Have experience with Olympiad level competitions (as a contestant, coach, or problem-setter)
Exposure to AI/LLM research, especially on evaluation or reasoning tasks.
Technical Setup: Desktop/Laptop with reliable internet connection and necessary software for mathematical computations, visual content creation, simulations, and online collaboration.
Job Classification
Industry: IT Services & ConsultingFunctional Area / Department: Research & DevelopmentRole Category: Research & Development - OtherRole: Post Doctoral ResearcherEmployement Type: Part time