I am a Ph.D. candidate in Computer Science at the New Jersey Institute of Technology. My research studies how LLM agents and distributed learning systems can reason over data, adapt to heterogeneous environments, and remain secure and trustworthy in practice.

LLM Agents LLM Safety Federated Learning

01

Education

New Jersey Institute of Technology
Ph.D. in Computer Science
Sep 2021 - Dec 2026 (Expected)
Newark, NJ
Khulna University of Engineering & Technology
B.Sc. in Computer Science and Engineering
Mar 2016 - Mar 2020
Bangladesh

02

Publications

Preprints

Peer-Reviewed Publications

03

Selected Projects

Current work01

Secure Agentic Data System

Developing a secure agentic data system that defends against prompt-injection attacks and enforces dynamic, fine-grained access control for query generation and execution.

LLM SafetyLLM AgentsAccess Control
Research project02

Agentic Text-to-SQL

Agentic Text-to-SQL system using schema exploration, multi-path SQL generation, SQL selection, correction rules, and query dependency trees.

GitHub Paper
ICLR 202603

Federated Learning for Heterogeneous Data

Clustered federated learning on heterogeneous non-IID data using data and gradient signals, with a dual-encoder architecture for privacy-preserving representation sharing.

GitHub Paper
AAAI 202304

Resource Allocation and Optimization

Equity-aware resource allocation formulated as an integer program and solved with weighted LP relaxation and randomized rounding.

Paper

04

Service

Reviewer: NeurIPS 2026, AAAI 2026, ACM CODASPY 2025, IEEE CIC 2025

05

Honors

Sigma Xi nomination, 2026; NJIT Graduate Student Assistantship; Codeforces Expert rating.

06

Technical Skills

LLM & Agents Retrieval-Augmented Generation, prompt engineering, MCP, tool calling, LLM evaluations

Fine-tuning PEFT/LoRA, SFT, DPO, ORPO, RLHF, PPO, GRPO, reward modeling

Systems PyTorch, CUDA, FSDP, DeepSpeed, vLLM, Slurm/HPC, Docker, Git, Linux

Languages and Data Python, SQL, C/C++, Java, Bash, JavaScript, Pandas, NumPy, PostgreSQL