Natural Language Processing (NLP) has significantly advanced language learning technologies, yet adaptive, personalized tools for less commonly taught languages like Japanese remain underdeveloped. Existing Japanese learning tools often lack dynamic adjustment to individual learners’ evolving proficiency, making the learning process daunting. Addressing this gap, we present RydeeNLP, a novel Japanese language learning system that integrates advanced NLP techniques with educational technology to provide personalized learning experiences. Our contributions in this paper are three-fold. First, we employ a word-swapping model to create a multi-tiered difficulty classification system for Japanese vocabulary, covering nouns, verbs, and adjectives. Second, we develop custom difficulty and paraphrase dictionaries to inform adaptive translation models that dynamically adjust translations based on learner proficiency. Third, we implement this technology in a web browser extension that provides real-time, proficiency-matched translations and automatically populates a spaced repetition flashcard system for ongoing learning. Evaluation results demonstrate the effectiveness of our classifiers and translation models, with the fine-tuned model producing more grammatically accurate translations and word-swapping model translating with overall higher BLEU scores.