This is your work, valued

United Arab Emirates

Noufal

Expert
@kvsnoufal

Principal Data Scientist @ Dubai Holding, UAE

portfolio-optimization. Python

104

reinforce. Implementing REINFORCE algorithm on Pong, Lunar Lander and Cartplot + Medium Article

23

Pytorch-FaceNet-DogDataset. Paper To Code implementation of Facenet in native pytorch on Dog Dataset

22

dashcam-object-detection. Label images with LabelImg; Object detection with detectron2

13

ImageGeoLocation. Building a neural network that predicts that can geotag an outdoor image and how to catch a cheating neural network with grad-cam

13

GauGanPytorch. Paper To Code implementation of NVIDIA's GauGan on a custom Landscape 's Dataset. Generating photorealistic-ish:p images from drawings

12

MSPhi2-for-classification-LoRA. Training Microsoft Phi 2 LLM for sequence classification with LoRA and QLoRA

7

webdevagent. Langchain Agent for Web Development With GPT-4o

7

MidiTransformer. GPT transformer from scratch, training to generate midi music , using Pytorch

6

Latent-Space-Interpolation. Using Autoencoders with keras and tensorflow.js

6

pix2pix. Paper To Code implementation of pix2pix GAN

4

signboard_text_detection. Generate synthetic dataset for object detection (Faster R-CNN)

3

alphazeroTicTacToe. Coding AlphaZero algorithm from scratch to play the game of TicTacToe and it never loses!!

3

neural_style_transfer. A CNN based approach to transfering "artistic style" from one image to another

3

finetunegemmacolorpalette. finetune gemma 370m for generating color palettes

2

langchainWorkshop. Jupyter Notebook

2

qwenvlattention. Visualizing how Qwen3-VL attends to image regions during caption generation

2

llm_sampling. Exploring Top p, greedy sampling and temperature on LLM generation

1

wordFaceNet. Python

1

flux-4bit-interpolation. flux-4bit for prompt interpolation

1

ambigramPytorch. Generating Ambigram with PyTorch

1

dubaistablediffusionvideo. Prompt interpolation and latent walk using Stable Diffusion.

1

gemma4uncensor. hands-on study of how to reduce model refusals (“uncensor” in the sense of weight edits, not retraining) for Gemma 4 E2B IT by estimating a per-layer refusal-related direction from activations and projecting it out of o_proj and down_proj while preserving row norms

1

dogvideodetector. Jupyter Notebook

1