Research
Work across time-series forecasting, retrieval-augmented systems, and trustworthy conversational fraud detection. I recently presented my ICLR 2026 poster. ICFD-31k is now published in the IJCAI-ECAI 2026 proceedings. ScopeBench-PR received Best Paper Presentation and placed runner-up in the GlobalSouthAI 3MT.
Research papers
Jul 21, 2026
1st place Best Paper Presentation Runner-up Three Minute Thesis (3MT)Do LLM Reviewers Respect Scope? ScopeBench-PR: A Benchmark for Scope Fairness in Peer Review
Presented at GlobalSouthAI (IJCAI-ECAI 2026) · GlobalSouthAI @ IJCAI-ECAI 2026
Presented at GlobalSouthAI @ IJCAI-ECAI 2026, earning 1st place for Best Paper Presentation and runner-up in 3MT. A counterfactual benchmark for measuring scope fairness in LLM peer review.
Apr 30, 2026
ICFD-31k: A Large-Scale Dataset and Benchmark for Real-Time Conversational Fraud Detection
IJCAI-ECAI 2026 · IJCAI-ECAI 2026 Special Track (Main Conference)
Published at IJCAI-ECAI 2026 (AI and Social Good). A large-scale Indian bilingual fraud-call benchmark with streaming labels and rationales.
Mar 2, 2026
Retrieval Mechanisms Surpass Long-Context Scaling in Time Series Forecasting
ICLR 2026 · 1st ICLR Workshop on Time Series in the Age of Large Models
Published at the ICLR 2026 TSALM Workshop. A paper arguing that selective retrieval beats brute-force long context in time-series forecasting.