Ludovic

Data Scientist

Abacus.AI • Freiberufler

Mar 2023 - Present • 3 yrs 2 mos

As a Senior Data Scientist and Consultant, I interacted with clients to understand their business needs and provide them with tailored A.I solutions: ⚡ NLP Discriminative Solutions: Developed NLP solutions, including named entity recognition, document classification, using transformers architectures. ⚡ Large Language Models: Utilized Foundation Models for various use cases such as - document generation (Articles, Legal, Reports, ...) - Question Answering (using RAG) - terms extraction - Chatbot - text analysis using Clustering and LLMs

Deep Learning/Machine Learning Engineer

Logically

Jan 2021 - Mar 2023 • 2 yrs 2 mos

As a Deep Learning engineer, I interacted with stakeholders to designed machine learning solutions to detect misinformation and threats: ⚡ Developed solutions for acquiring signals from various data sources, including OCR and transcription for English and multilingual content. ⚡ Developed end-to-end solutions, from dataset creation to training, benchmarking, and monitoring using MLFlow. ⚡ Implemented simple deployment using REST APIs (Flask, FastAPI) and Gradio for stakeholder demos. ⚡ Developed solutions for object detection (weapons), landmark recognition (geolocation), and emotion recognition using PyTorch and TensorFlow. Several solutions have been deployed to facilitate fact-checking of content from diverse sources, including images and video, on a daily basis. Additionally, specialized risk-detection tools have been implemented for use during specific events.

AI Engineer

Crisp, a Kroll business

Jan 2019 - Jan 2021 • 2 yrs

As a A.I engineer, I designed end to end solutions to detect risks on social media platforms: ⚡ Developed and deployed deep learning solutions for risk detection on social media platforms. ⚡ Pretrained and fine-tuned models (CNN, BERT, Electra, RoBerta) for text and video data (sentiment analysis, video classification, risk classification). ⚡ Served models using TorchServe and deployed them on Google Cloud Platform (GCP) using Docker and Google AI Platform. ⚡ Pushed the limits of deep learning models through strategies like Knowledge Distillation and Ensembling for smaller yet high-performance models. Models have been deployed, allowing moderator to filter hundreds harmful contents per day.