I am the Founder of Code Studio and the Principal Lead at Code Studio AI Research Lab. With over 5+ years of experience leading cross-functional engineering teams, building scalable mobile/web platforms, and deploying production AI systems, I am now directing my core focus toward academic research and open-source innovations in Deep Learning, HCI, and AI4Science.
- 📍 Based in Bangladesh
- 🔬 Research Focus: Low-Resource NLP, Machine Learning for Science, Trustworthy AI, & Bio-Computing
- 🏢 Operating Code Studio (Active ecosystem of 20+ successfully shipped web/mobile applications)
- 🎓 Academic Trajectory: Preparing for upcoming Graduate/PhD research applications in the United States
Founded and operational for over 5+ years, managing a dynamic, high-performing team of 18+ software engineers, designers, and systems architects. We architect high-barrier applications, including enterprise e-commerce infrastructures, secure medical documentation frameworks, and robust consumer automation grids.
An active, open-source research initiative dedicated to solving complex, real-world social and scientific crises through deep learning. We bridge abstract theoretical computational intelligence with cross-disciplinary hardware/software systems co-design.
Architecting fail-safe systems for highly critical domains, ensuring factual alignment, strict citation verification, and linguistic adaptability.
- TrustMed-RAG-Framework: Fact-checking and self-correcting RAG network built explicitly for bulletproof clinical support setups.
- Clinical-Banglish-Robustness: Core benchmarking tools measuring language model resilience against heavy code-mixed medical token sequences.
Expanding the frontiers of machine learning to capture and translate biological time-series data and molecular coordinates.
- Plant-BioSignal-Translation: Translating plants' micro-voltage electrical action potentials into natural language stress notifications.
- Olfactory-Language-Model: Utilizing Graph Neural Networks (GNNs) to digitize 3D molecular structures into continuous odor vector coordinates.
-
Bio-Processing-Units-BPU: Compiling binary algorithms into nucleotide paths (
$A, T, C, G$ ) for zero-waste, soil-biodegradable computing simulations.
Pioneering alternative input interfaces and custom speech architectures to redefine human-machine communication loops.
- Neuromorphic-Code-Compiler: Transformer pipelines compiling non-invasive raw EEG brainwave sequences directly into backend source code syntax.
- BanglaASR-Dialect-Wav2Vec: Finetuned acoustic processing networks recognizing low-resource dialects and regional language distributions.
| Category | Domain Technologies |
|---|---|
| AI / Machine Learning | PyTorch, PyTorch Geometric, Hugging Face, Scikit-Learn, MNE-Python, RDKit |
| Software Architecture | Node.js, Flutter, Dart, Python, JavaScript, REST/GraphQL APIs, CI/CD Pipelines |
| Research Artifacts | LaTeX, Core Abstract Syntax Trees (AST), Signal Filtering, Time-Series Forecasting |
- LinkedIn: linkedin.com/in/soykot-hosen
- Hugging Face: huggingface.co/CodeStudioAIResearchLab
- Research Lab: Code Studio AI Research Lab
- Website: code-studio4.com
💡 Let's build a smarter, safer, and ecologically accountable future through computational intelligence.