MURSALEEN HASSAN

COMPUTER VISION ENGINEER

I am a driven Software Engineer specializing in Computer

Vision. My expertise lies in leveraging advanced algorithms CONTACT ME AT
and techniques to solve complex visual recognition problems.
I am deeply passionate about pushing the boundaries of +923320357371
computer vision through innovative research and application.
mursaleenhassan24@gmail.com
WORK EXPERIENCE
LinkedIn Github
Intermediate Python Engineer
Entrolics LLC | Oct 2024 - Present Karachi, Pakistan
Developed Python microservices and deployed them on Kubernetes
clusters in Google Cloud Platform (GCP).
Created end-to-end Vertex AI applications using GCP. PROJECTS
Implemented robust error handling for Python code.
Shelf Products Counts
Integrated microservices with front-end Node.js applications.
Problem: Counting each shelf product by category via a mobile app.
Implemented Yolov8 detection model trained on a SKU dataset to count products on
Senior Computer Vision Engineer the shelf.
Employed PyTorch OpenCLIP ViT model embeddings for classification.
The Disrupt Labs | Oct 2021 - 2024 Stored embeddings in MySQL using Django ORM.
Utilized the Faiss library for fast product similarity computation.
Implemented REST APIs for efficiency.
Developed an end-to-end Python pipeline with Django ORM database, Deployed the project on a GCP GPU-integrated instance with a reserved IP.
incorporating machine learning, deep learning, and classical OpenCV. Trained the model on Google Colab.
Utilized machine vision-based cameras. Integrated endpoints for front-
end communication.
Fabric Defect Detection
Deployed projects on AWS and GCP platforms, utilizing GPU and non-GPU Problem: Detecting defects at a speed of 1 meter/second.
Utilized multiple machine vision cameras covering a width of 3.5 meters and
instances. Managed resources such as reserved and elastic IP addresses,
concatenated multiple camera images via multithreading.
S3 buckets, and cloud databases (SQL, MySQL, PostgreSQL). Utilized classical and deep learning approaches to achieve anomaly detection at
Collaborated with the Web and UX Team, connecting the backend with 160 frames/second.
Stored defects in MySQL using Django ORM.
frontend via Web Sockets, Rest APIs, and Databases. Developed a PyQT5 desktop application to display defects to end-users.
Collaborated closely with web development and UX teams to integrate Generated PDF reports using open-source Python libraries.
backend functionality with frontend interfaces using Web Sockets, REST
APIs, and databases.
Factory Monitoring
Created UML diagrams and developed SQL databases for storing various Problem: Monitoring crowds, pallets, wrong parking, and emergency
data types, including person anomaly alerts, crowd alerts, fabric and exits.
Utilized detection models to detect persons, vehicles, and pallets.
catalyst defects, measurements, and OCR data. Developed algorithms to send crowd, wrong parking, and pallet alerts via WhatsApp
in specific regions of interest.
Implemented a PyQT5 desktop application to display KPIs, matrices, anomalies, and
alerts.

EDUCATIONAL HISTORY
Managed 70 cameras simultaneously using multiprocessing and multithreading.

Box OCR
Sir Syed University of Engineering and Technology Problem: Counting boxes and detecting Lot IDs on conveyors.
BS IN SOFTWARE ENGINEERING | 2020 - 2024 Employed detection algorithms to detect boxes on conveyors and stop conveyors
via PLC.
BMJ Digital Education Utilized machine vision cameras to detect Lot IDs with over 97% accuracy.
Coordinated communication between PLC and machine vision cameras for
Computer Vision Course | Aug 2021 - Oct 2021 simultaneous image capture.
Utilized Nvidia GPU 3060 for enhanced speed.
NAVTTC - From NED University
Big Data Analytic Technique | Sep 2020 - Mar 2021 Catalyst Anomaly Detection
Problem: Detecting anomalies in moving large amounts of catalyst on
Aptech - Computer Education conveyors.
Advanced Diploma In Software Engineering | 2015 - 2018 Utilized multiple machine vision cameras to capture images.
Employed deep learning anomaly detection with over 95% accuracy.
Managed light and wiper triggers via Python and machine vision cameras.

SKILLS SUMMARY
Saved anomalies in MySQL using Django ORM.
Developed a PyQT5 interface for reporting and live production monitoring.
Implemented one-click PDF reports for specified date ranges.
Developed a runtime training algorithm to handle various catalyst categories.

Programming and Coding: Python Sent email alerts for anomalies and machine statuses.

Embedded: Raspberry Pi Jetson Nano T-Shirt and Socks Measurement

Problem: Measuring T-shirts and socks from various perspectives.
Machine Learning and Computer Vision: Utilized machine vision cameras and controlled lighting via pulse.
Employed OpenCV classical approaches to detect socks and T-shirts via contours
Tensorflow Pytorch Scikit-Learn Pandas Numpy Efficientnet after image preprocessing.
YOLO Zero Shot Detection PaddleOCR EasyOCR CraftOCR Utilized NumPy for measuring width, length, and other dimensions.
Achieved a measurement error rate of 0.5 inches.
Anomalib Segment-Anything Roboflow Labelimg
Documents OCR
Deployment Plateforms: AWS GCP Problem: Extracting structured data from PDFs or images.
Utilized PaddleOCR, EasyOCR, and CraftOCR for extraction.

Frontend and Backend: PyQT5 HTML5 Flask Rest API’s Preferred CraftOCR for its polygon and square box detection capabilities.
Achieved over 90% accuracy with PaddleOCR for recognition.
Employed PaddlePaddle for structure analysis.
Sockets Django ORM Streamlit Utilized AWS Textract and Google Vision for high accuracy.
Employed OpenCV for image preprocessing.
Operating System: Windows Linux Achieved 98% accuracy in national identity card extraction (Pakistan, Bangladesh,
Nepal, and Sri Lanka).

Software Project Management: Github Jira Clickup Achieved over 90% accuracy in extracting data from 120+ bank cheques across
multiple countries (Pakistan, Bangladesh, Nepal, and Sri Lanka).

Database: SQL MySQL