Koshik Debanath
I’m a Software Engineer at Universal Machine Inc., where I build AI-powered applications and decentralized governance platforms. Previously, I worked as a Data Scientist at Manaknightdigital Inc., developing end-to-end ML solutions including RAG systems, chatbots, fraud detection, and custom image generation platforms.
My research spans three interconnected areas: natural language processing, generative AI, and deep learning. Each builds upon lessons from the previous, driven by a central question: How can we make the latest advancements in visual computing more accurate and effective, particularly in healthcare settings?
I earned my BSc in Computer Science & Engineering from Rajshahi University of Engineering & Technology (RUET) in 2024. Currently, I’m a Research Member at Young Learners’ Research Lab (YLRL), where I conduct interdisciplinary research in NLP, Generative AI, and misinformation detection. My work spans computer vision, time series forecasting, and business analytics. Recently, I published research on low-resource NLP for Bengali language, AI-generated text classification, and physics-informed neural networks.
Research Interests:
- Natural Language Processing (NLP)
- Generative AI (LLMs, RAG, Fine-tuning)
- Deep Learning and Computer Vision
- Explainable AI (XAI)
- Physics-Informed Neural Networks
- Misinformation Detection
news
| July, 2025 | Submitted journal article on "Bayesian Physics-Informed Neural Networks for Parameter Inference and Uncertainty Quantification in Reaction-Diffusion Models of Wound Healing" to Mathematical Biosciences (Under Review). Preprint available. |
|---|---|
| April, 2025 | Joined Universal Machine Inc. as Software Engineer I, developing AI-powered applications, Web3 governance platforms, and blockchain-based DAO systems. |
| March, 2025 | Paper on "Analyzing Bot Activity and Political Discourse in the 2024 U.S. Presidential Election" published in NCIM-2025. |
| March, 2025 | Contributed to OpenLLMetry open-source project - Fixed Python dataclass serialization bug in workflows (PR |
| February, 2025 | Paper on "Distinguishing Human-Written and AI-Generated Text using XAI" published in NCIM-2025 (co-author). |
| February, 2025 | Paper on "Physics-Informed Neural Networks for Real-Time Anomaly Detection in Power System Dynamics" accepted at BIM 2025. |
| January, 2025 | Paper on "Advancing Low-Resource NLP for Bengali Language Using Llama" published in ECCE-2025. |
| January, 2025 | Paper on "Distinguishing Between Formal and Colloquial Bengali" published in ECCE-2025 (co-author). |
| November, 2024 | Completed "Understanding and Applying Text Embeddings" course from DeepLearning.AI. |
| November, 2024 | Contributed to OpenLLMetry - Fixed TypeError in OpenAI embeddings metrics handler (PR |