Python libraries (pandas, trimesh, plotly, meshlab, PyTorch Geometric) plus 3D and simulation workflows: mesh autoencoders, SDF, and FEM data.
About
I am an AI Solutions Engineer with a PhD in Artificial Intelligence, specializing in turning real-world, cross-disciplinary challenges into practical and reliable AI systems. My journey into AI began in my twenties, before frameworks such as TensorFlow and PyTorch were available, when I designed and trained a neural network for my BSc thesis in Software Engineering. Since then, I have worked across industrial computer vision, data-centric AI, time-series analysis, 3D deep learning, machine learning for FEM simulations, LLMs, VLMs, RAG, AI agents, and end-to-end AI development—primarily with complex, real-world data rather than standard benchmarks. Having witnessed AI evolve from low-level implementations to today’s AI-assisted development, I remain deeply fascinated by its progress and possibilities.
PhD in Artificial Intelligence
Technical University of Chemnitz
Applied AI Focus
Industrial computer vision, data-centric AI, and production-oriented workflows
Applied & Scientific Foundation
Practical AI for real-world systems, grounded in scientific machine learning and research methods
Core Expertise
Computer Vision
Object detection, instance segmentation, industrial image understanding, field-condition evaluation, and visual inspection workflows.
Data-Centric AI
Dataset design, annotation strategy, class balancing, failure-case analysis, and iterative model improvement.
AI Systems
End-to-end pipelines, workflow integration, production-oriented design, validation logic, and deployment planning.
Multimodal AI
Vision-language workflows: semantic retrieval, document/PDF understanding, and structured extraction from mixed media.
Deployment
CoreML, Swift/iOS integration, on-device inference, model export, and mobile visualization prototypes.
Scientific ML
FEM surrogate modeling, 3D deformation learning, mesh autoencoders, implicit representations, and RL for process modeling.
Skills
PyTorch, TensorFlow, and scikit-learn for model development, plus computer vision: detection, segmentation, industrial imagery, preprocessing, and evaluation.
Frameworks
Docker and Podman environments, conda, Linux, Git, Jenkins, Jupyter, n8n automation, Hermes-based local model workflows, multimodal RAG, and document/PDF understanding.
MLOps & Workflow Tools
OOP and languages including C#.NET, C++, MATLAB, and Python; deployment with CoreML, Swift/iOS, on-device inference, and model export.
Programming & Deployment
AI-assisted engineering workflows for prototyping, coding, debugging, and research acceleration with modern editor and agent tooling.
AI Development Tools
Research & Education
PhD in Artificial Intelligence
Technical University of Chemnitz, Germany | 2026
Thesis: Exploring Deep Learning Approaches for 3D Deformation: Toward Finite Element Method Distillation
M.Sc. Mechatronics Engineering
K.N. Toosi University of Technology, Iran | 2010
Thesis: Visual Servoing and Object Pose Estimation - 6 DOF Robot Manipulator
B.Sc. Computer Software Engineering
IAU Tehran Branch, Iran | 2006
Thesis: Online Persian Handwriting Recognition using Neural Networks
Professional Experience
AI Solutions Engineer - TKI mbH, Germany
2024 - Present- Develop AI solutions for industrial inspection, infrastructure analysis, document understanding, and semantic retrieval.
- Design computer vision systems for object detection, duct segmentation, color detection, and noisy real-world field imagery.
- Prepare deployment-oriented prototypes, including CoreML export and iOS visualization workflows.
- Improve model performance through data-centric AI: dataset refinement, annotation strategy, class balancing, and failure-case analysis.
Research Scientist - Technische Universitaet Chemnitz, Germany
2018 - 2024Research career focused on machine learning for engineering simulation, 3D deformation modeling, automotive body production, and environmental sensor analysis across BMBF-funded applied research projects.
ML@Karoprod: Machine Learning for Automotive Body Production
BMBF project with Fraunhofer IWU, SCALE GmbH, and TU Chemnitz.
- Created large-scale FEM training datasets for sheet metal forming where no suitable dataset existed.
- Used implicit neural representations, including signed distance functions, for dense deformable geometries.
- Developed surrogate models for geometric deviation, thickness distribution, and thinning on high-resolution shell meshes, including the ML-Karoprod MeshPredictor.
- Built reinforcement learning approaches for inverse process design and sequential deformation modeling.
Smart Airsense: AI-Based Health Assistant
BMBF project with Corant GmbH / air-Q and TU Chemnitz.
- Worked with air-quality sensor measurements logged over time by Corant / air-Q devices.
- Added machine learning methods for predicting environmental events from noisy real-world time-series data.
- Developed and validated predictive models in a human-in-the-loop machine learning setting.
Selected Applied Projects
Industrial Vision: Field Detection and Deployment Prototypes
Applied AI · 2024 - PresentBuilt field-ready computer vision workflows for construction-site and infrastructure imagery.
- Designed segmentation pipelines for trenches, ducts, fibers, and fittings in construction-site imagery.
- Handled lighting variation, occlusion, dirt, shadows, and ambiguous material boundaries during evaluation.
- Converted models to CoreML and built iPhone prototypes for on-device visualization and testing.
- Connected annotation strategy, model evaluation, failure analysis, and deployment preparation in one workflow.
Local LLM, Multimodal Document Understanding and Semantic Search
Applied AI · 2024 - 2026Developed privacy-conscious local AI workflows for documents, images, and knowledge extraction.
- Built local LLM-based workflows for document understanding, semantic search, and knowledge extraction without external cloud inference.
- Extracted structured information from complex technical documents, PDFs, and tables.
- Combined retrieval, vision-language models, and RAG-style approaches for automated document analysis.
- Built CLIP-based semantic image search using natural language queries, embeddings, and similarity-based clustering.