Khadim H.

ML Engineer | Data Scientist

elunic AG • Tiempo completo

Aug 2022 - Present • 3 yrs 10 mos

Implemented and trained segmentation and detection models for Industry 4.0. • Developed and fine-tuned segmentation and detection models using PyTorch and TensorFlow, deploying them with Docker and Kubernetes for efficient model management in an MLOps environment. • Implemented an API server to facilitate model inference, handling multiple requests efficiently, while utilizing Git for version control and collaboration. • Optimized inference speed through the integration of TensorRT, ONNX, and Triton, resulting in a 70% improvement over previous systems. • Monitored feedback post-deployment, utilizing a dynamic feedback loop to iteratively enhance model performance. • Achieved an impressive accuracy rate of 99.999% in most cases, significantly improving company model performance from 70% to 99.999%, a gain of 29.999%. • Conducted research on new models and advancements to stay current with the latest technologies, contributing to ongoing innovation within the team. • Led data annotation efforts and managed a team to ensure high-quality annotated datasets for model training

LLM Developer

Jamie • Freelance

Feb 2022 - Dec 2023 • 1 yr 10 mos

• Collected and annotated a large text dataset for an LLM project, implementing models for text classification to generate answers based on questions and context. • Developed efficient data collection and annotation pipelines to streamline the preparation of datasets for LLM model training.

Computer Vision Engineer | Data Scientist

Education • Freelance

Jan 2022 - Dec 2023 • 1 yr 11 mos

• Built end-to-end TensorFlow models for computer vision and LLM, optimizing training pipelines and conducting performance evaluations to iteratively refine models, ensuring they meet key metrics and deliver optimal results. • Engineered inference pipelines, achieving 90% faster predictions and enhancing deployment efficiency.