Experience

Applied machine learning, research engineering, education, and service.

Professional experience

Vector InstituteTorontoSep 2025 — present

Associate Applied Machine Learning Specialist

  • Built an automated benchmark-generation system grounded in textbooks and technical references, including OCR ingestion, multi-agent task design, verification, iterative repair, and semantic deduplication.
  • Owned a large-scale medical figure extraction and preprocessing pipeline covering compound figures, subfigures, captions, and embedded subcaptions.
  • Reduced Qwen2.5-VL batched inference latency from 4.0s to 1.2s per sample and multimodal summary generation from 4.0s to 0.6s.
  • Supported OpenCLIP training on 18M image–text pairs through scalable preprocessing, quality checks, and reproducible workflow documentation.
Intelligent Visualization LabYork University · TorontoSep 2023 — Aug 2025

Graduate Research Assistant

  • Built reusable multimodal ML pipelines for dataset construction, preprocessing, experimentation, evaluation, and error analysis.
  • Developed agentic workflows and benchmarks for data storytelling, chart reasoning, and interactive dashboard question answering.
  • Co-developed GRPO post-training for text-to-visualization using post-execution multimodal feedback.
  • Evaluated open and closed VLMs and compact judges across accuracy, robustness, consistency, and bias protocols.
Islamic University of TechnologyBangladeshJul 2021 — Aug 2023

Lecturer

  • Designed and delivered theory and lab courses in structured programming, object-oriented programming, and systems analysis.
  • Mentored students in debugging, reliable software construction, and technical communication.

Education

York University2023 — 2025

MSc in Computer Science

CGPA: 8.6/9.0

Islamic University of Technology2016 — 2019

BSc in Computer Science and Engineering

CGPA: 3.77/4.0

Technical toolkit

Core engineering

Python, Linux, Git, SQL, Slurm, modular systems, profiling, and distributed workflows.

ML and multimodal systems

PyTorch, Hugging Face, vLLM, DeepSpeed, Vertex AI, and multimodal LLM/VLM workflows.

Post-training and evaluation

SFT, DPO, GRPO, reward modeling, model-as-judge, robustness evaluation, and error analysis.

Leadership and service

Aug 2024 — Aug 2025

President, Bangladeshi Graduate Student Association at York

Jan 2023 — present

Peer reviewer

NeurIPS, ACL, EMNLP, COLM, NAACL, and ACL Rolling Review.

Outside work, I enjoy technical and general-interest podcasts, reading articles, and playing video games.