Muhammad Nauman Amjad

Academic

Verbatim text

Muhammad Nauman Amjad
~
+923145767753, nouman.amjad.mail@gmail.com
~
1-10, Islamabad.
Linkedln: http_s:LLwww.linkedin.comlinLmuhammad-nouman-amjadL
Github: btti:i:i:llgitbub.tQm[l'llQumaa-Amiad
Education
Bachelor of Science (Computer Science)
I 2021-2025
Major: I
Generative Al, Computer Vision, MLOps, Machine Learning, Deep Learning, Blockchain, Cloud Computing, Data Structures,
Algorithms, Obiect-Oriented Programming
Projects
Finni Proiect: Geo-Tagging US Roadway Assets Using 2D Images [Pytorch, Tensorflow, Opencv, Flask, GeoPy, React]
Developed a deep learning based system to accurately geotag roadway assets (e.g., traffic signs) using 2D images. The aim is to enhance
asset management for US transportation agencies bv minimizing geolocation error, achieving accuracy within a 7-feet margin.
Portfolio Proiects:
Al-Powered Personal Memory & Knowledge Assistant [Al, NlP, Vector Search] :
- Developed an Al agent that stores, retrieves, and connects knowledge from documents, meetings, and videos, improving information
ecall for students, researchers, and professionals.
'\PfflP<:tPr PrniPrtc·
Synthetic Data Generation for Enhanced Object Detection of Roadway Assets [GANs, Transformers, and Diffusion Models] :
- Developing a synthetic data generation pipeline with the goal of reducing detection error by 60%, significantly enhancing the accuracy of
object detection for roadway assets.
- Generating t housands of synthetic images using GANs, Transformers, and Diffusion Models, augmenting t he KITII dataset to improve
training diversity, and achieving a notable increase in Precision (mAP) metrics.
Automatic Georeferencing System using local Feature-based Image Matching [local Feature Extraction, Homography, RANSAC]:
- Developed a system to georeference UAV images by mapping pixel locations to a reference satellite image. Implemented SIFT, SURF for
feature ext raction and RANSAC for homography estimation.
- Enabled real-time retrieval of geodetic coordinates from aerial images using pixel projection and interpolation.
Automated Attendance Tracking [YOlOvB, FaceNet, Pytorch, Tensorflow, Flask, Bootstrap] :
- Developed a face recognition attendance system using YOLOv8 & FaceNet for attendance tracking saving 20% time of lecture.
- Built with Flask and Bootstrap, featuring student management, attendance history, and cross-device accessibility.
Pollut ion Prediction System (Mlflow, Prometheus, Grafana]:
- Developed a real-time environmental monitoring and pollut ion prediction system using modern MLOps tools. The project included
automated data collect ion with the OpenWeat herMap API, a predictive LSTM model deployed with Flask, and real-t ime system
monitoring using Prometheus and Grafana.
Random Forest Classifier With DVC Pipeline [Python, Ml, DVC) :
- Designed a scalable and reproducible DVC pipeline to t rain and evaluate a Random Forest Classifier.
- Automated stages for data preparation, feature extraction, model t raining, and evaluation. Reproducible results using version-controlled
data and models. Evaluation metrics with visualization of feature importance and ROC curves.
Work Experience
Ml Engineer, iEngneering, Islamabad. [Project-Based]
il\pr 2024-
Collaborating with transconomyto develop a deep learning-based system for geolocating US roadway assets using 2D
May 2025
mages. Enhanced geolocation accuracy by over SO%, driving error reduction from 17 feet to under 7 feet using
vehicle-mounted footage.
Skills & Tools
Professional Skills
lnteroersonal Skills Problem-Solvine: Teamwork Time Manae:ement Flexibilitv. Adaotive.
Technical Skills
Python, AWS, NumPy, Pandas, PyTorch, TensorFlow, Scikit-Learn, Matplotlib, Seaborn, Flask, Docker, Kubernetes,
Jenkins, MIFlow, DVC ,Prometheus ,Grafana, SQL, MongoDB, GO, Java, C++ HTML, CSS, React, Bootstrap.
Achievements
Dean List Certificate [Spring 24, Fall 24).
Trainings I Workshops
TinyML Workshop [Deploying Al Models on Edge devices]
Interests
Playing and watching Football, Find something new to learn
FAST NUCES ISLAMABAD CAMPUS

AI enrichment