Research & Applications

Research topics, framed as transferable capabilities.

I do not want this blog to be boxed into one thesis topic. These categories keep the research side broad: multilingual systems, evaluation, responsible AI, retrieval, and applied ML engineering.

Research & application tracks

These labels are designed to cover current work and leave room for future AI/NLP directions.

Multilingual AI

Machine translation, multilingual NLP, low-resource adaptation, LoRA domain NMT, domain terminology, language variation, and cross-lingual evaluation.

Model Evaluation

Metric design, benchmark construction, paired comparisons, statistical testing, ablations, error analysis, and human evaluation.

Explainability & Responsible AI

Bias evaluation, interpretability, training-data attribution, Shapley analysis, controlled prompts, model auditing, and responsible data practices.

Retrieval & Knowledge Systems

RAG, search, sparse/dense/hybrid retrieval, graph retrieval, long documents, evidence grounding, and document QA.

Applied ML Systems

Data analysis, classification, forecasting, information extraction, ML pipelines, deployment, demos, and practical AI products.