Mohan KC

I’m passionate about Natural Language Processing and Trustworthy AI, especially when it comes to advancing the formal methods mindset behind language understanding and generation. I enjoy turning research ideas into working systems that people can actually use.

As a Junior Machine Learning Engineer, I developed chat analysis and spam detection models for live-streaming moderation in romanized Nepali text using fine-tuned XLM-RoBERTa and HuggingFace Transformers. I also built modules for live engagement and performance analysis, including time series forecasting with LSTM and viewer behavior classification with XGBoost using pandas and PyTorch. On top of modeling, I designed backend APIs for the live streaming app with FastAPI.

During my Python Developer Intern role, I contributed to backend development with Django and PostgreSQL. I integrated Generative AI capabilities into existing product pipelines using LangChain and OpenAI APIs (chat and whisper), and I built automation scripts to streamline internal workflows using selenium and celery.

My interests show up across my projects and publications—summarization with BART, speech emotion detection with Conformer-XGBoost, and RAG-based customer support systems. I’m motivated by building trustworthy, language-focused solutions end-to-end, from data and modeling to deployment.