LinkedIn
GitHub
Hi, I'm Isteyak.
I build intelligent systems that bridge algorithms and real-world impact.
As a recent Computer Science graduate from the University of Windsor, I've worked on deep learning–based computer vision models for geospatial analysis, optimized AI pipelines using PyTorch and TensorFlow, and contributed to research in image similarity, attention architectures, and mixed-precision training.
Now, I'm diving deeper into hardware-aware AI—exploring FPGA-based acceleration, embedded vision systems, and efficient deployment for applications like autonomous vehicles and industrial automation.
Beyond code, I bring discipline from industrial roles, clarity from teaching 200+ students, and a relentless drive to solve problems that matter—whether in the lab, the factory floor, or the cloud.
This is where I share my projects, experiments, and journey at the intersection of AI, systems, and practical engineering.
Skills
- Programming: Python, Java, C, C++, JavaScript, TypeScript, Rust, SQL, NoSQL, Bash/Shell, HTML, CSS, JSON, YAML, Linux/Unix, OOP, Functional Programming, Algorithms, Data Structures, Multithreading, API Development
- Machine Learning & AI: Deep Learning, Machine Learning, Artificial Intelligence, Computer Vision, CNNs, RNNs, LSTMs, Autoencoders, Siamese Networks, Transformers, Vision Transformers, GANs, Object Detection, Face Detection, Landmark Detection, Image Segmentation, Classification, Regression, Clustering, Embeddings, Representation Learning, Feature Engineering, Model Optimization, Model Deployment, Hyperparameter Tuning, Model Evaluation, Data Preprocessing, Data Augmentation, End-to-End ML Pipelines, MLOps, GPU Acceleration, Distributed Training
- Frameworks & Libraries: TensorFlow, Keras, PyTorch, OpenCV, scikit-learn, YOLOv5, YOLOv7, YOLOv8, YOLO11, RetinaFace, MTCNN, Dlib, MediaPipe, Transformers (HuggingFace), NumPy, Pandas, Matplotlib, Seaborn, SciPy, FastAI, ONNX, OpenMMLab, GDAL, Rasterio, ArcGIS API, QGIS API
- Web & Full-Stack: React, Next.js, Node.js, Express.js, MongoDB, Mongoose, Supabase, PostgreSQL, REST APIs, GraphQL, WebSockets, JWT Authentication, OAuth2, Server-Side Rendering (SSR), Static Site Generation (SSG), Microservices, Serverless Functions, Tailwind CSS, Vite, Redux, API Integration, JSON APIs, MVC Architecture
- DevOps & Tools: Docker, Kubernetes (K8s), Git, GitHub Actions, CI/CD, Linux, Bash, Nginx, PM2, Load Balancing, Logging & Monitoring, SSH, Version Control, Containerization, Virtual Environments, Testing & Debugging, Unit Testing, Integration Testing
- GIS / Geospatial: Remote Sensing, Satellite Imagery, Raster Processing, Geospatial Analysis, Spatial Data, Georeferencing, Coordinate Reference Systems (CRS), Image Tiling, Patch Extraction, GIS Pipelines, QGIS, ArcGIS, Earth Observation Data, GeoTIFF Processing
- Cloud & Platforms: AWS (EC2, S3, Lambda), Google Cloud (GCP), Firebase, Vercel, Netlify, Railway, Heroku, Colab, Kaggle, Cloud Deployment, Cloud Functions, Serverless Architecture
- Other: Data Structures & Algorithms, Object-Oriented Design (OOD), Software Design Patterns, Microservices, System Design, API Design, Software Architecture, Performance Optimization, Scalability, Debugging, Code Review, Agile, SCRUM, Technical Documentation
Education
Bachelor of Computer Science
University of Windsor | Windsor, ON | January 2023 - December 2024
- Transferred Credits from Independent University, Bangladesh (January 2021 – August 2022)
- Earned Dean's Awards for Fall 2021 and Winter 2022
Professional Experience
Researcher | GeoVision Lab, University of Windsor | Windsor, ON | February 2025 - Present
- Developed a patching system for creating splits of any satellite image and tested using QGIS render tech.
- Created a web scraping system to extract data in a structured way from database websites.
- Trained a Siamese Attention-UNet model using TensorFlow/Keras on geospatial image pairs to detect visual similarity for change detection tasks.
- Optimized model training on limited GPU resources (NVIDIA GTX 1080) by tuning batch size, input resolution, and memory growth settings to avoid OOM errors during large-scale inference.
Research Assistant | University of Windsor | Windsor, ON | October 2024 - December 2024
- Researched and evaluated frontal face detection techniques by integrating YOLOv8n with MTCNN and YOLO11 with RetinaFace, achieving improved detecting accuracy and frame-level face counts in ML experiments.
- Conducted performance benchmarking and model comparison to assess inference speed, scalability, and detection reliability for computer vision applications.
- Collaborated under the supervision of Dr. Ziad Kobti, Former President of the Canadian AI Association, contributing to innovative approaches and automating development workflows for experiments.
Teaching Assistant | University of Windsor | Windsor, ON | August 2023 - December 2024
- Assisted over 200 students in developing a strong foundation in software engineering concepts such as Data Structures and Algorithms (DSA), providing guidance on problem-solving techniques, building performant solutions, and writing secure, maintainable code.
Projects
- Contributed to building an immersive 3D classroom in Three.js with interactive areas for collaborative learning.
- Integrated Cohere's Command A Vision Model and latest LLM to process handwritten notes, achieving rapid LaTeX conversion and generating polished study materials in 3 seconds.
- Used autoencoder to embed each image into a single vector for feature extraction and clustering.
- Applied KNN algorithm to cluster similar faces and the accuracy of the heuristic is recorded for evaluation.
- Assisted in API integration using React, Geocodify, and Electricity Map for real-time data visualization.
- Collaborated in a team of 4 to develop a resource solution, securing 3rd place in the competition.
Leadership Experience
ACM Club Event Assistant | Independent University, Bangladesh | Dhaka | March 2021 - August 2022
- Assisted with organizing 10 events per annum related to computer science and technology with no less than 70% attendance per event.
- Collaborated with others as a team for event logistics, including venue selection, setup, and equipment arrangement, ensuring events were well-prepared and visually appealing.
Additional Projects
- Orchestrated Docker containers for a Python web app, ensuring seamless deployment across various environments and platforms.
- Utilized Docker for isolating a Binary Search Tree implementation in Python, optimizing resource utilization.
- Implemented a robust CI/CD workflow using GitHub Actions, automating the build and deployment processes for both the web app and Binary Search Tree applications.
- Designed the About Page, enhancing UX with interactive live backgrounds.
- Assisted in integrating React with Geocodify and Electricity Map APIs for real-time data.
- Worked in a team of 4 to create electrical usage resource solutions for potential clients and placed 3rd in the competition.
- Automated machine-independent digital resume web page portfolio based on Streamlit technology.
- Cleaned a database and converted it to a JSON file using Streamlit technology.
- Implemented a search bar option for user-friendly data retrieval.
- Automated machine-independent digital resume web page portfolio based on Streamlit technology.