AceGuardian

Game solving researcher(CUDA, C++) (Remote)

Remote
Work Type: Contract

The Role


We are seeking a researcher to drive forward our work in game theory, developing proofs of concept, reviewing literature, and implementing cutting-edge research. Ideal candidates include PhD students in game theory or authors of relevant papers. While deep engineering expertise is not required, candidates must demonstrate proficiency in C++ and Python.


Your responsibilities:

  • Design and develop advanced game-solving algorithms with a focus on optimizing Counterfactual Regret Minimization (CFR) for large-scale imperfect information games.

  • Implement and fine-tune CFR variants, including Monte Carlo CFR (MCCFR), vectorized CFR, to improve computational efficiency, quality of the solutions and convergence speed.

  • Apply reinforcement learning and search-based techniques to enhance game-solving strategies and opponent modeling.

  • Analyze and improve the performance of game-solving AI by implementing regret minimization strategies, function approximation methods, and neural network-assisted CFR techniques.

  • Collaborate with AI researchers and game engineers to integrate game-solving AI into production systems and evaluate real-world performance.


What you bring to the table:

  • 3+ years of experience in game theory, reinforcement learning, or AI for strategic decision-making.

  • Expertise in CFR algorithms and their optimizations, including variants like MCCFR, vectorized CFR, depth-limited CFR, and regret matching techniques.

  • Deep understanding of reinforcement learning, including value-based and policy-based methods.

  • Familiarity with search-based planning methods such as Monte Carlo Tree Search (MCTS).

  • Strong mathematical foundation in probability, statistics, optimization, and game theory.


Preferred qualifications:

  • PhD or Master’s in Computer Science, Machine Learning, AI, Mathematics, or a related field.

  • Proven track record of research publications in game theory, reinforcement learning, or multi-agent systems.

  • Strong analytical skills and a collaborative mindset.

  • Ability to write in Python and C++


Nice to have:

  • Experience with Poker, solvers, and solution libraries

  • Experience with CUDA

  • 2+ years of commercial experience

  • 3+ years of C++ experience    

Submit Your Application

You have successfully applied
  • You have errors in applying