AI/ML @ TU Darmstadt · Robotics & AI · GenAI systems that actually ship

_{sub-average github profile}

🌐 Website · 🎓 Google Scholar · 💼 LinkedIn

• 🎓 M.Sc. Artificial Intelligence & Machine Learning @ TU Darmstadt — deep generative models, multimodal AI, NLP... professionally convincing matrices to have opinions
• 🦾 TA @ PEARL Lab (Intelligent Robotic Manipulation) — designing the lab assignments where 61 students and one very patient Franka Panda figure out perception → planning → control together
• 🧠 AI Developer @ BLACKLAKE, Frankfurt — co-leading the AI department, shipping RAG pipelines and agentic workflows (n8n, Power Automate, or models trained from scratch) so companies' documents finally talk back
• 📄 Main-authored a LiDAR odometry paper at ACRA 2024 — we benchmarked loop closure in an actual Australian botanic garden 🌿 (eucalyptus trees: surprisingly good landmarks)

computer, show em where we've been.

graph LR
    A["🎓 Bremen<br/>B.Sc. Robotics (Hons)"] --> B["🖐️ DFKI<br/>robot hands in MuJoCo"]
    B --> C["🌊 Porto<br/>LIO-SAM + octrees"]
    C --> D["🦘 Sydney<br/>LiDAR odometry → ACRA'24"]
    D --> E["🧠 Darmstadt<br/>M.Sc. AI/ML + Franka TA"]
    E --> F["🏦 Frankfurt<br/>RAG in production"]

_{Trajectory drift: accumulating · Loop closure: pending}

• Australian Centre for Robotics, Sydney 🇦🇺 — main-authored "Performance Assessment of LiDAR Odometry Frameworks: A Case Study with Loop Closure at Australian Botanic Garden Mount Annan" (ACRA 2024)
• Ingeniarius, Porto 🇵🇹 — built a LIO-SAM pipeline turning LiDAR point clouds into octrees to spot novel environmental info fast. Big maps, small patience.
• DFKI, Bremen 🇩🇪 — simulated the anthropomorphic MIA Hand in MuJoCo for the EU APRIL project, and measured exactly how hard robot fingers squeeze (they should know their own strength)

• iRobMan_Franka — Franka Panda perception–planning–control sims & tooling, straight from the PEARL Lab course I TA. The same assignments 61 students dream about (mostly at 3 a.m.)
• COIN-LIO (fork) — notes & experiments on intensity-augmented LiDAR-inertial odometry, for when geometry alone gets lost in long corridors
• MarketOps_DBorse — pipeline + evaluation for finance-ops use cases. The "GenAI that actually ships" department 📈

The brains 🧠

The muscles 🦾

The glue 🔧

• 🗣️ Languages: English, French & Arabic (fluent) · German (C1) · Python (native) · C++ (it's complicated)
• 🏆 2× DAAD RISE Worldwide Scholar (2023 & 2024) — Germany kept sending me abroad to play with robots
• 🎪 Co-ran jacobsHack!22, one of Central Europe's biggest hackathons — turns out herding 5 departments is harder than herding robots

📧 mourad.ouazghire@stud.tu-darmstadt.de · or DM on LinkedIn

_{🛰️ README last calibrated: 2026 · sensor drift within acceptable bounds}