Muneeba Nasir
Machine Learning Engineer @CNRS
Signup · Get unlimited contacts
WORK HISTORY
Machine Learning Engineer @CNRS
Provence-Alpes-Côte d\'Azur, FR
Developing end to end ML pipelines in Python for large scale multimodal datasets (images, audio, video, text); GPU accelerated via CUDA and PyTorch with Bash automated multiprocessing.Integrated Prismer Large (vision language model) for semantic scene description at scale; validated and analysed outputs across large visual corpora.Systematically benchmarked multiple deep learning architectures (RetinaFace, MediaPipe, Mask2Former, UniDet): designed structured evaluation protocols.Integrating ShotBench, a VLM for cinematographic comprehension, via custom prompt engineering to extract cinematic features from video data; training a downstream classifier on the extracted representations.Fine tuned a graph neural network for action classification from pose keypoints; iterated through ablation studies and quantitative evaluation cycles.Conducted rigorous statistical hypothesis testing (ANOVA, permutation-based p-value tests, baseline comparisons) to validate model performance with statistical rigour.
EDUCATION
National University of Computer and Emerging Sciences
Bachelor of Science, Computer Science
Université de Bordeaux
Master's degree, Neuroscience
ABOUT MUNEEBA NASIR
Machine Learning Engineer with 4+ years of experience specialising in LLM evaluation frameworks, model benchmarking, and production AI pipelines. At CNRS I3S, I develop GPU accelerated multimodal pipelines, benchmark Vision Language model architectures, and fine tune models with structured evaluation and ablation studies. On the applied side, I built and shipped an end to end document intelligence product using fine tuned LLM, owning evaluation pipelines that improved extraction accuracy by 25%. I bring direct experience with healthcare and medical NLP and a strong track record of translating evaluation insights into measurable product improvements.
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.