À propos de Mohammed Shabbir
Anglais
Bilingue ou natif
Français
Capacité professionnelle limitée
Arabe
Notions
Hindi
Bilingue ou natif
Ourdou
Bilingue ou natif
Expériences
- ZEMOTECHFreelance AI & ML EngineerCENTRES DE RECHERCHEmai 2024 - Aujourd'hui (2 ans et 2 mois)Paris, France○ Independent AI/ML practice delivering GenAI, computer-vision, and evaluation systems to R&D clients end to end — scoping,build, evaluation, and hand-off.○ Built and shipped public and client AI projects across RAG, QLoRA fine-tuning, LLMOps regression gates, and validated NLP screening (see AI Highlights above).○ Developed a graph-attention-network (GNN) anomaly / spoof detector on a 120,798-sample adversarial dataset, achievingPR-AUC 0.9401 and F10.9070 under strict leave-one-attack-out evaluation — transferable to fraud, intrusion, and securityanalytics.○ Delivered a 14-month R&D contract (VEDECOM) building hybrid test/evaluation frameworks and real-time monitoring dashboards
- VEDECOM InstituteR&D AI EngineerTÉLÉCOMMUNICATIONSjuin 2019 - décembre 2024 (5 ans et 6 mois)78000 Versailles, France○ Developed a receiver-side misbehavior detection approach for V2X / CPM data using vehicle perception context, published at IEEE VTC Fall 2023 2 Online .○ Built machine-learning regression and clustering models to estimate vehicle speed and distinguish pedestrians, bicycles, and vehicles from Bluetooth sensor and ground-truth data.○ Created Bokeh-based analytics dashboards and KPI pipelines for real-time monitoring, daily and weekly traffic patterns, and congestion analysis from roadside sensing data.○ Contributed to cooperative perception simulation and V2X integration work using OMNeT++, Veins, SUMO, CPM/CAM processing,and scenario-based evaluation.
- FlowareChief Technology OfficerTRANSPORTSmars 2022 - mars 2023 (1 an)Paris, France○ Led technical development of a roadside traffic-sensing platform combining computer vision and Bluetooth-based analytics fortransport monitoring.○ Deployed YOLO v4-v7 object detectors and SORT, DeepSORT, and OCSORT trackers on Nvidia Jetson Nano, Raspberry Pi, and Google Coral TPU for real-time transport-mode detection and tracking.○ Developed track-recovery and post-processing algorithms that improved counting and tracking accuracy by 95% againstmanually prepared ground truth, delivering the end-to-end pipeline from edge inference to time-series KPIs.
Recommandations
Ces profils de freelance correspondent également à vos critères
Agatha Frydrych
Backend Java Software Engineer
4.7
(3)
2
Baptiste Duhen
Fullstack developer
4.6
(4)
5
Amed Hamou
Senior Lead Developer
4
(2)
7
Audrey Champion
Web developer
4.3
(3)
4
Formations
- PhD in Computer Science and NetworksParisTech2017Computer Science and Networks
Certifications
- PhD CertificateTelecom Paris2017