The official research wing of Code Studio, dedicated to pushing the boundaries of artificial intelligence, machine learning, and natural language processing to solve complex societal challenges. Our core focus centers on sustainability, reliability, low-resource NLP, and healthcare innovations.
We drive technical innovation and academic publications across four pillars:
- Green & Sustainable AI: Architecting energy-efficient Transformer networks, dynamic model pruning, and sparsification frameworks to minimize carbon footprints in deep learning.
- Trustworthy & Explainable AI (XAI): Breaking open neural "black boxes" through mechanistic interpretability layers to eliminate hallucinations in critical text generations.
- Low-Resource Speech Processing: Developing robust Automatic Speech Recognition (ASR) pipelines for regional dialects and underrepresented languages.
- AI for Science & Multimodal Healthcare: Accelerating drug discovery via generative molecular modeling and crafting multi-modal data fusion grids for ultra-early disease detection.
Our open-source framework ecosystem is segmented into specialized modules:
- TrustMed-RAG: Verification-driven RAG architecture for clinical support.
- Clinical-Banglish-Robustness: Stress-testing text models under heavy code-mixed distributions.
- BanglaASR-Dialect-Wav2Vec: Fine-tuned Meta's Wav2Vec 2.0 for regional dialects.
- MultiModal-Alzheimers-Detection: Cross-attention data fusion engine for early-stage neurodiagnosis.
💡 Let's build a smarter, safer, and more sustainable future together.
