Looking for a job

Ismaïl Mostefaoui

@ismalmostefaoui

Message

Data scientist building Python models, computer vision pipelines, and AI tools after engineering at BNP Paribas Fortis.

Belgium

Message

What I'm looking for

I closed my venture in May 2026 to refocus on data science and AI.

At BNP Paribas Fortis, I migrated a legacy document archiving system to a full-stack search engine handling billions of documents in a regulated environment. I also optimized APIs and batch processing under GDPR constraints.

At MostPharma, I built a B2B marketplace for independent pharmacists and reached 100 users. I closed the venture in May 2026 to refocus on data science and AI.

In my data science projects, I built scikit-learn pipelines to predict positive COVID-19 cases, reaching an F1 score of 0.60 and recall of 0.75 on positive cases. For my MSc thesis, I benchmarked genomic sequence compression tools, with reference-based compression reaching a 330x ratio.

At ACIC, I benchmarked CNN face detectors and selected YOLOv3, which ran at nearly 30 FPS on CPU. I also prototyped a face-tracking and multi-camera re-identification pipeline, and I'm currently deepening my data science skills at ULB.

Experience

Work history, roles, and key accomplishments

Founder and Technical Lead

MostPharma

Nov 2025 - May 2026 (6 months)

Built end-to-end in 6 months a B2B marketplace for independent pharmacists, reaching 100 users. Led market research, pricing, and sales.

Next.js TypeScript Supabase Market Research Pricing Sales

Software Engineer

BNP Paribas Fortis

Sep 2021 - Dec 2024 (3 years 3 months)

Migrated a legacy document archiving system to a full-stack search engine handling billions of documents in a regulated environment. Optimized APIs and batch processing, applying TDD and Jenkins CI/CD under GDPR constraints.

React Java Spring MVC Test Driven Development Jenkins CI CD GDPR API Optimization Batch Processing

AI / Computer Vision Intern

ACIC

Sep 2020 - Nov 2020 (2 months)

Benchmarked CNN face detectors (YOLOv3 vs SSD-ResNet) and selected YOLOv3 at nearly 30 FPS on CPU. Built a SORT-inspired tracking pipeline with IoU and face embeddings, prototyping multi-camera re-identification.

YOLOv3 SSD ResNet CNN OpenCV TensorFlow Sort IOU Facial Embeddings Cross Camera Re Identification

Education

Degrees, certifications, and relevant coursework

Université libre de Bruxelles

Training, Data Science

2026 -

Ongoing training in Data Science Applied to Health & Life Sciences, covering Python, R, SQL, statistical modelling, and data visualization.

Machine Learnia

Training, Machine Learning

2026 - 2026

Completed training on Machine Learning with Scikit-learn, covering NumPy, pandas, seaborn, classification, ensemble learning, clustering, anomaly detection, and PCA.