Machine Learning Engineer based in Berlin, Germany with expertise in LLMs, MLOps, and software engineering. Experienced in building and fine-tuning large language models, creating embedding models, and developing cloud-native ML platforms. ICPC World Finalist and passionate problem solver with a strong background in algorithmic thinking.

• LLM Engineering: Fine-tuning, evaluation, embedding models
• Frameworks: PyTorch, TensorFlow, Transformers, TRL
• ML Tools: DVC, Argilla, MT-Bench
• Vector DBs: Qdrant, Weaviate

• Languages: Python, Java, C/C++, JavaScript, Golang, TypeScript, R
• Backend: Django, Flask, FastAPI, NodeJS, Express
• Cloud: AWS, GCP
• DevOps: Docker, Kubernetes, Terraform, CI/CD (GitHub Actions, GitLab CI)
• Databases: PostgreSQL, MySQL, MongoDB

• Contributed to training jina-embeddings-v3 with 8k sequence length and multi-lingual capabilities (2M+ downloads/month on HuggingFace)
• Enhanced Starcoder language model through distillation and synthetic data training, achieving up to 27% pass@1 on HumanEval
• Built Jina AI Cloud, a cloud-native MLOps platform for multimodal AI and neural search
• Maintained open source MLOps framework (Jina) and vector DB library (DocArray)

• Contributed to the release of the Jina 3.0 cloud-native MLOps framework
• Added search engine components based on AI models to Jina hub
• Assisted in building Jina AI Cloud with AI and search services

• Improved entity recognition modules, ML models, and web crawlers
• Managed AWS infrastructure for data processing services
• Led team to achieve 97% overall product accuracy

AI-powered Terraform generator that creates cloud resources using natural language.

Features:

• Natural language-based resource creation
• Support for AWS cloud resources (S3, EC2, etc.)
• Local infrastructure development with LocalStack
• Component-based infrastructure management
• Self-healing infrastructure creation

Tech Stack: Python, Terraform, AWS, LocalStack, OpenAI

AI-powered website creator using LLMs and VLMs to create, render and edit websites.

Demo Video

Previous Version (2021):

• E-commerce and portfolio website builder
• Backend: NodeJS, Express, MongoDB
• Infrastructure: AWS ECS, S3, Route53
• CI/CD with GitHub Actions and Terraform

Web application using AI models to classify medical documents into ICD10 categories.

Tech Stack:

• Pre-trained BERT model for document encoding
• Classification model built with PyTorch
• Backend: Python and Django
• Infrastructure: AWS ElasticBeanstalk with CI/CD

• ICPC World Finalist (2021) - First Tunisian team to qualify for this prestigious programming competition
• First Prize OpenGovDataHack (2020) - MaintainIt: dashboard for predictive maintenance to reduce flight delays
• Maghreb Champions, ACPC (2019) - Arab and Africa Collegiate Programming Championship
• AWS Developer Associate Certification (Feb 2020 - Feb 2022)

Software Engineering Degree
National Institute of Applied Science and Technology (INSAT)
Sep 2017 - Jun 2022, Tunis, Tunisia

• English: Full Professional Proficiency
• French: Full Professional Proficiency
• Arabic: Native or Bilingual Proficiency

Open to collaboration and new opportunities in AI and ML engineering.