Experience

Research, teaching, and engineering across the full experimental lifecycle.

From problem formulation and reproducible implementation to evaluation, visualization, manuscript preparation, teaching, and AI-agent assessment.

Education

July 2024 – Present

North South University · Dhaka, Bangladesh

M.S. in Computer Science and Engineering

GPA: 3.55/4.00; 24 of 30 credits completed. Supervisor: Dr. Mohammad Ashrafuzzaman Khan.

Thesis: Ultra-Efficient Fine-Tuning of 7B-Parameter Large Language Models under Constrained VRAM using Dynamic Progressive Compression and Adaptive Stride Training.

May 2019 – June 2023

North South University · Dhaka, Bangladesh

B.S. in Computer Science and Engineering

GPA: 3.53/4.00; graduated Cum Laude.

Thesis: BERT Knowledge Distillation for NLP Tasks, later extended into a 2025 PLOS ONE article.

Research

2024 – Present

Department of Electrical and Computer Engineering · NSU

M.S. Thesis Researcher

Built a single-GPU fine-tuning framework for Qwen2.5-7B-Instruct.
Combined 4-bit loading, QLoRA, gradient checkpointing, adaptive rank scheduling, SVD warm-start, and GPU isolation.
Conducted 27 controlled experiments measuring quality, memory, runtime, convergence stability, and schedule sensitivity.
Engineered pinned environments, controlled random seeds, configuration logging, and per-experiment tracking.

2023 – Present

Efficient NLP Group · North South University

Research Collaborator

Conduct research with Dr. Mohammad Ashrafuzzaman Khan on knowledge distillation, efficient NLP, and experimental validation.
Contributed to peer-reviewed work on Transformer compression and computationally efficient misinformation detection.
Participated in problem formulation, experimental design, statistical validation, visualization, scientific writing, and camera-ready preparation.

Professional

Nov 2025 – Present

MindriftAI · Remote

AI Evaluation and Pilot Task Contributor

Design, review, and validate evaluation scenarios for autonomous AI agents.
Assess logical soundness, behavioral alignment, realism, and policy adherence in multi-step workflows.
Improve model behavior through prompt refinement, curated exemplars, structured feedback, and iterative validation.

Dec 2022 – Oct 2023

North South University

Teaching Assistant

Conducted tutorials and office hours for undergraduate computer-science courses.
Prepared course materials, assignments, grading rubrics, and detailed feedback.
Explained technical concepts to mixed-ability cohorts and supported curriculum development.

Technical depth

Methods and tools

Machine Learning & NLP

PyTorch, Hugging Face Transformers, scikit-learn, TensorFlow, JAX/Flax (basic), spaCy, BERT, RoBERTa, GPT-family models, Mistral, Qwen, RAG, LangChain, and vector databases.

Model Efficiency

Knowledge distillation, 4-bit quantization, LoRA/QLoRA, gradient checkpointing, adaptive rank scheduling, pruning, efficient evaluation, and GPU-memory profiling.

Research Methods

Controlled baselines, ablation studies, matched comparisons, leakage prevention, significance testing, null models, cross-validation, F1/ROC-AUC, and PRISMA-guided synthesis.

Systems & Tooling

Python, SQL, PHP, CUDA, Linux, Git/GitHub, reproducible environments, LaTeX, Docker concepts, n8n, Tableau, Microsoft Office, data structures, algorithms, and scripting.