<p>Healthcare accessibility remains a significant challenge for Tamil-speaking communities due to the lack of effective AI-driven communication tools that support Tamil-English code-mixed speech. Developing a reliable Automatic Speech Recognition (ASR) system for Indian languages is particularly difficult due to the limited availability of large-scale, high-quality speech datasets. This challenge becomes even more pronounced in noisy, code-mixed environments with diverse grapheme vocabularies. Similarly, the development of robust Question Answering (QA) systems for Tamil faces two major obstacles: resource scarcity and linguistic complexity. Existing Tamil QA datasets are small-scale, increasing the risk of overfitting, while dedicated corpora for Tamil-English code-mixed healthcare QA are nonexistent. This paper presents an AI-Healthcare Assistant integrating ASR, QA, and Text-to-Speech (TTS) technologies to improve healthcare accessibility for Tamil-speaking users, particularly in code-mixed settings. It proposes a novel ASR system tailored for low-resource, code-mixed Indian languages to enhance speech recognition and information retrieval in real-world healthcare applications. A key innovation of this work is the strategic incorporation of data synthesis and knowledge distillation, which significantly enhance model performance despite data scarcity. Experimental results show that the fine-tuned Whisper model achieves a substantial improvement in code-mixed speech recognition, reducing the Word Error Rate (WER) from an initial baseline of 1.016 to 0.54, achieving an approximate 50% relative reduction. Additionally, XLM-RoBERTa, benefiting from data synthesis and knowledge distillation, attains a BLEU score of 0.4000, outperforming other models in Tamil-English question answering. By leveraging these techniques, this work effectively bridges linguistic gaps, improving AI-driven healthcare support for Tamil-speaking communities.</p>