01.
AI 2048: Hybrid RL + Expectimax Search
Cracking 2048: A Hybrid Approach using PPO, C++ Expectimax, and Bayesian Optimization
SEC 01 // HERO
ALEJANDRO CASTRO
AI RESEARCHER & SYSTEMS ENGINEERING STUDENT. FOCUSING ON REINFORCEMENT LEARNING, ALGORITHMS, AND SEARCH-BASED PLANNING.
SEC 02 // RECENT_WORK
Projects
02.
Rikal
Archived local-first flashcard application combining FSRS scheduling, offline-capable synchronization, AI-assisted card generation, and semantic answer evaluation.
03.
FABRIK Inverse Kinematics
Interactive p5.js implementation of the FABRIK inverse-kinematics algorithm for a variable-length articulated arm following a user-controlled target.
SEC 03 // RECENT_BLOGS
Blogs
Beyond Heuristics: Solving 2048 with Hybrid Reinforcement Learning and C++ Expectimax Search
How a hybrid agent combines masked PPO, C++ expectimax search, Bayesian optimization, and batched ev…