Work
I started in AI in 2021, while still in engineering school: internships in Zurich and with NTT DATA Munich alongside my degree (Esprit, 2025), then full-time from 2025. Today I'm AI Team Lead at 1morething Studio and I teach working professionals in the evenings; in summer 2026 I also built RL agent-evaluation tasks for Anthropic's frontier models at Originator. Every number below comes from the project's own benchmarks or production data.
Where I've shipped
Experience
AI Team Lead
- Own the AI roadmap for every product in the 1MT venture portfolio: I validate architecture and model decisions, set the studio's AI service standards, and review the AI services the team ships. Mentor 2–4 engineers.
- Absar, edge vision for a coffee chain in Riyadh: built all 11 on-device hygiene and service checks now live on NVIDIA Jetson. A GPU benchmark I ran made inference 17× faster; retraining lifted cup-detector precision from 0.875 to 0.992.
- SheGrows, for FAO in Egypt: built the Egyptian-dialect retrieval layer and the expert feedback loop of a WhatsApp RAG assistant that has answered 15K+ farmer questions.
- LegalTech, for a ministry: built an on-premise pipeline for scanned Arabic legal files. Cut OCR time per line 45× with identical output.
Present● Live
AI Evaluation Engineer · Freelance
- Built ~40 agentic evaluation tasks that test frontier models, Anthropic's Claude Sonnet and Opus, against real client software, so those clients can ship agents built on the models. ~80% acceptance rate.
- Calibrated computer-use RL (RLVR) tasks to a 20–80% pass band across three difficulty tiers: compound-step difficulty design, small-sample confidence bounds before any verdict, and run budgets that absorb ~30% infra failures. Deterministic Python graders on the app's final state.
- Validated tasks for Qwen3 for a second client.
Jul 2026
Data Science Instructor
- Teach the AI Full Pack, Python through Transformers, end to end: 3 cohorts, ~50 students, mostly working professionals.
- Classes run in English, French, and Tunisian depending on the room. Hands-on, project-based.
Present● Live
Data Scientist
- Built a QHSE translation service in Spring AI + Groq covering 10 languages, with per-tenant caching and parallel batch translation. I then exposed it as an A2A agent and wrote a Spring AI MCP server on top of it.
- Also shipped the in-app assistant and audit-automation agents. 301 automated tests and a 9-stage GitLab CI with security scanning.
Feb 2026
Data Scientist
- Classified 100K+ parliamentary speeches into topics. Pipeline hit 95%.
- Built a RAG system so citizens could ask questions about what was said in parliament.
Sep 2025
AI Engineer · Internship
- Embedded straight into the NTT DATA team in Munich. Led the RAG research and chose the LLM on AWS SageMaker.
- Built the cloud pipelines too: S3, EC2, EKS, Docker, GitLab CI.
Dec 2024
Data Scientist · Internship
- Trained a RoBERTa pipeline on 1M+ German legal records. Got +15% downstream accuracy.
- Automated the processing side: 40% faster end to end.
Oct 2023
Projects
Absar: Edge Vision for a Coffee Chain
SheGrows: Arabic RAG for FAO Egypt
LegalTech: Arabic Legal Archives
Ahki: Tunisian Arabizi → Arabic
DocMate: AI Documentation Assistant
RAG and MCP with Spring AI
Personal projects
Qirat: Crypto Decision Desk
Source on GitHub. Also: Voyagent, a CrewAI multi-agent trip planner I audited and hardened (safe calculator tool, grounded prompts, Groq retries).
Activity
Reviews
Abdelkerim has a good and solid understanding of RAG systems and deploying LLMs on AWS, and approached his work with curiosity and a structured mindset. He was proactive in solving challenges and took the time to deepen his understanding when needed.
Your clear and practical way of teaching complex topics (from Python fundamentals to machine learning, deep learning, Generative AI, and Transformers) made the entire learning journey truly enriching. Your passion, patience, and continuous support greatly strengthened my understanding.
Your real-world examples and clear guidance revealed the incredible potential of AI. We truly appreciate your patience as you led us through hands-on projects, deep-learning fundamentals, and encouraged us to explore beyond the basics. You've inspired us to apply these skills in real-world projects.
J'ai particulièrement apprécié sa grande motivation et sa capacité à apprendre rapidement. Il est une personne autonome et sérieuse. Je suis convaincue qu'il réussira dans sa future carrière.