Sameer lmran Ahmed
FAST
· 2025
Email
sameerimranahmed@gmail.com
Phone
+923215411414
LinkedIn
—
GitHub
—
Academic
Program
—
CGPA
—
Year
2025
Education
—
Address
—
DOB
—
Verbatim text
The exact text the LLM saw on the page (or the booklet text from the old import).
This is what powers semantic search.
Sameer lmran Ahmed
+923215411414, sameerimranahmed@gmail.com
E-11, Islamabad.
Linkedln: www.linkedin.com/in/sameer-imran-7 48b01284
Github: https://github.com/sameer-imran
eep Learning, Machine Learning, Computer Vision, NLP, Generative Al, Cloud Computing,
mage Processing.
he City School, Capital Campus Islamabad
-levels (Mathematics, Computer Science, Physics, Chemistry)
he City School, Capital Campus Islamabad
inal Pro ·ect: INS{P)ECT - Optimized Flea Beetle Counting using Edge Devices {R&D)
•
In collaboration with the University of Regina, and Agriculture and AgriFood Canada.
•
Compared various models (YOLOll, FasterRNN, Retina Net, RE-DETR) for insect counting on sticky traps.
•
Different variations of YOLO (nano, small, large), different backbones with FasterRCNN (Inception ResNet,
EfficientNet).
•
Utilized two datasets to train and validate models: an open-source public dataset and a proprietary dataset.
•
Researched ways to improve inference time and mAP.
•
Develo ed a mobile a
lication in Flutter and API for the o timized model to be hosted on AWS.
emester Pro ·ects:
ideo Frame Prediction using Deep Learning Models on UCF101 Dataset [PyTorch, OpenCV, Python]
•
Developed and compared ConvLSTM, PredRNN, and Transformer models for video frame prediction using
the UCF101 dataset.
•
Evaluated model performance using metrics like MSE and SSIM, achieving MSE < 0.02 and SSIM > 0.8.
rdu to English Machine Translation Using Attention-Based Encoder-Decoder Model [PyTorch, Python]
•
Developed a custom RNN-based attention mechanism for a neural machine translation model to translate
Urdu to English using PyTorch.
•
Experimented with different word embeddings, and evaluated the model on BLEU score.
net-Based Automated Glaucoma Detection [PyTorch, OpenCV, Python]
•
Developed a segmentation model using U-Net architecture for optic disc (OD) and optic cup (OC) in retinal
fundus images.
•
Achieved >90% segmentation accuracy/sensitivity, meeting glaucoma diagnosis standards.
ampusCompass: University Portal [C#, MySQL, WinForms]
•
Built a university portal in Windows Forms providing student, instructor, and admin functionality.
•
Features include course management, assignment upload and submission, automatic grading, and
attendance trackin .
kills & Tools
rofessional Skills
roblem-Solving, Critical Thinking.
echnical Skills
ython, C/C++, C#, Java, .NET, PyTorch, TensorFlow, SQL, Sci-kit Learn, CNN, RNN, Transformers,
chievements
warded full scholarship on tuition fee in A-Levels for achieving straight A*/As in 0 levels.
*AAA across four A-levels.
rainings I Certification
WS Certified Cloud Practitioner - (AWS Academy)
nterests
hess Badminton C clin
Video Games.
enMPI.
FAST NUCES ISLAMABAD CAMPUS
AI enrichment
Sameer Imran Ahmed is a recent graduate from FAST NUCES with a focus on Deep Learning, Computer Vision, and NLP. He has practical experience through academic and research projects involving model optimization, mobile application development, and cloud deployment on AWS.
Skills (AI)
["Python", "PyTorch", "TensorFlow", "Computer Vision", "Deep Learning", "NLP", "AWS", "Flutter", "C#", "SQL", "YOLO", "U-Net", "Transformers"]
Status: ai_done
Provenance
Source file: —Created: 1777724106