GESA: Exploring Loss-based Adversarial Attacks in Volumetric Media Streaming
IEEE MIPR 2024
GESA dataset for packet-loss impacts on streaming point clouds.
š¤ About
I am a Machine Learning expert specializing in R&D of LLMs and Multimodal LLMs (MLLMs) that integrate text, images, and video. I design scalable transformer architectures for cross-modal reasoning and alignment. My work spans fine-tuning, optimization, and deployment of LLMs/MLLMs for captioning, VQA, video-language grounding, and document intelligence.
š„ News
- šš
2025.07: Served as Ethics reviewer at NeurIPS 2025 Datasets and Benchmarks Track!
- šš
2024.08: GESA has been selected at IEEE 7th International Conference on Multimedia Information Processing and Retrieval (MIPR)!
- šš
2022.05: Lightseg has been accepted at IEEE 19th International Symposium on Biomedical Imaging (ISBI)!
š Publications
LIGHTSEG: Efficient Yet Effective Medical Image Segmentation
IEEE ISBI 2022
Lightweight segmentation for constrained compute.
Pay Attention for COVID-19 Detection Using Efficient Convolution
IEEE ISBI 2022
COVIDAT-Net: low-compute COVID-19 classification.
š Education
Toronto Metropolitan University
Toronto, Canada
Degree: MEng, Electrical & Computer Engineering ā CGPA: 3.63 / 4.33
Relevant Coursework:
- Neural Information Processing
- Machine Learning
- Topics in Data Science
- Deep Learning
Rajshahi University of Engineering & Technology (RUET)
Rajshahi, Bangladesh
Degree: BSc, Electrical & Electronic Engineering ā CGPA: 3.62 / 4.0
Relevant Coursework:
- Introduction to Programming Language
- Engineering Mathematics IāV
- Electromagnetic Fields & Waves
- Introduction to Digital System & Design
- Microprocessor & Microcomputer System
- Advanced Computer Programming
š Certifications
- How Google does Machine Learning ā Google Cloud Training (Coursera)
- Launching into Machine Learning ā Google Cloud Training (Coursera)
- Convolutional Neural Networks in TensorFlow (Coursera)
- Introduction to TensorFlow for AI, Machine Learning, and Deep Learning (Coursera)
- Introduction to Containers w/ Docker, Kubernetes & OpenShift (Coursera)
- Containers & Kubernetes Essentials (IBM)
- Design Thinking for Innovation ā University of Virginia (Coursera)
š¼ Experience
Concordia University
Research Assistant
- Conducted research in 3D vision, point cloud video compression, and 3D video data processing.
- Developed and optimized novel algorithms for point cloud completion and enhanced 3D video processing for real-time streaming.
- Published and presented findings at academic conferences, contributing to advancements in 3D real-time video streaming.
Tools: Focus Areas: Point Cloud Completion, Point Cloud Video Compression
Jan 2023 ā Dec 2024 | Montreal, Quebec
CINTIQS
Senior Artificial Intelligence Engineer
- Achieved significant advancements in AI innovation for military and defense applications by developing end-to-end AI software pipelines. Successfully trained and deployed new AI models for OCR, video super-resolution, and object tracking. Designed and implemented image restoration techniques using GANs for denoising, colorizing historical images, and frame interpolation, enhancing the quality of old and degraded videos.
- Implemented services using Flask, OCR, and Docker.
Tools: Python, Flask, SQLite, OCR
January 2022 ā Sep 2022 | Ottawa, Ontario, Canada
Atelesys
Software Developer
- Developed end-to-end applications using Python and React to efficiently meet project requirements.
Tools: Python, Flask, React
Oct 2021 ā Dec 2022 | Toronto, Ontario, Canada
Intelense
Artificial Intelligence Developer
- Developed real-time video analytics for public safety: anomaly detection (GAN, WGAN, VAE), accident and fall detection (pose estimation), fight, fire and smoke detection.
- Integrated AI with real-time camera feeds (RTSP, HTTP) using deep learning and computer vision.
- Built modular Flask + OpenCV apps for multi-camera tracking (perspective transform, object detection, tracking).
- Conducted R&D, literature reviews, and built pipelines across projects.
- Implemented alert systems to notify on anomaly threshold breaches.
Tools: Python, Flask, JavaScript
Jun 2020 ā Sep 2021 | Toronto, Canada
Dutch-Bangla Bank
Data Analyst
- Conducted data mining and retrieval with MySQL to identify critical business areas; identified top customers using clustering.
- Delivered a POC solution achieving 96% accuracy on company data.
Tools: Python, MySQL
Jun 2010 ā Apr 2016 | Dhaka, Bangladesh
š Awards
- Split Graduate Fellowship GCS ā 2023
- Concordia Conference and Exposition Allowance ā 2024
š¤ Service
- Volunteer, Toronto AI Meetup Group
- Organizer, University AI Symposium 2023
š ļø Technical Skills
- Programming Languages: 6+ years with Python (PyTorch, TensorFlow, Keras), C/C++, R, SQL, MATLAB.
- LLM & Generative AI Expertise: Fine-tuning & prompt engineering for LLaMA, GPT-3/4, Mistral, Gemma. Hugging Face + LoRA/QLoRA.
- AI Model Training: LLM, MLLM, CNN, 3D CNN, RNN/LSTM/GRU, Object Detection, OpenCV, U-Net, GAN.
- Computer Vision: Real-time analytics, video mining, 3D detection, anomaly detection, super-resolution (SRGAN).
- AI Deployment & Optimization: AWS SageMaker, Flask, REST API.
- Model Compression & MLOps: TensorRT, TensorFlow Lite, Git, OpenShift, Docker.
- Data Management & Big Data: Spark-Scala, Hive, Flume, Sqoop, Pig, Databricks, PostgreSQL, MongoDB.
- Fullstack & Web: React, HTML, CSS, RESTful APIs; integrate backend AI into responsive UIs.
- Operating Systems & Cloud: Mac, Windows, Linux, ROS; AWS (EC2, S3, EMR, SageMaker, Lambda).
- GPU Computing: NVIDIA GPU, CUDA, cuDNN, Keras, PyTorch, TensorFlow, TensorBoard.