Summary
G Shahariar is a PhD student and researcher specializing in Trustworthy AI and Large Reasoning Models, currently at UC Riverside with a perfect 4.00 CGPA and publications at EMNLP 2024 and ICLR 2026 on safety, adversarial attacks, and multimodal alignment. Over a decade of experience spans academic teaching and leadership, industry ML engineering, and research contributions in low-resource multilingual NLP and API analysis (ASE 2023). He has built practical datasets and state-of-the-art systems—like the Bengali fake-review dataset (BFRD) and an ensemble BanglaBERT solution—and applied SFT, DPO, adversarial training, and interpretability methods in real projects. Former ACM ICPC regional contestant turned mentor, he has guided students on RAG and bias mitigation while shipping deployed QA/summary models on SageMaker and performance-critical C++ components for VR/AR. He combines deep theoretical work on hierarchical reasoning with hands-on engineering across Python, C++, and PyTorch, making models both more capable and more reliable.
10 years of coding experience
3 years of employment as a software developer
HSC, Science, A+, HSC, Science, A+ at Dhaka College
Doctor of Philosophy - PhD, Computer Science, 4.00 out of 4.00, Doctor of Philosophy - PhD, Computer Science, 4.00 out of 4.00 at University of California, Riverside
Bachelor of Science - BS, Computer Science, 3.973 out of 4.00, Bachelor of Science - BS, Computer Science, 3.973 out of 4.00 at Ahsanullah University of Science and Technology
SSC, Science, A+, SSC, Science, A+ at Ideal School and College