Ali Bina
Verified Expert in Engineering
Machine Learning Engineer and Developer
Düsseldorf, North Rhine-Westphalia, Germany
Toptal member since August 26, 2025
Ali is a results-driven ML engineer with a PhD and 10+ years of experience deploying cutting-edge AI solutions. An expert in designing and implementing machine learning models, he has a strong background in Python, PyTorch, MLOps, and LLMOps architectures and a passion for leveraging AI to improve operational efficiency and outcomes. Ali is adept at collaborating with multidisciplinary teams to deliver impactful AI-driven solutions and translating business needs into scalable technology.
Portfolio
Experience
- Deep Learning - 10 years
- PyTorch - 10 years
- Azure - 8 years
- Large Language Models (LLMs) - 4 years
- Hugging Face - 4 years
- Azure AI Search - 4 years
- Open-source LLMs - 3 years
- Agentic AI - 2 years
Preferred Environment
Azure, Azure Machine Learning, GitHub Codespaces, Docker
The most amazing...
...thing I've developed is a multi-agent large language model (LLM) platform for autonomous materials discovery, blending AI, HPC, and knowledge graphs.
Work Experience
Senior AI/ML Engineer
Microsoft
- Architected and deployed a multi-agent LLM system for autonomous materials discovery, integrating patent search, knowledge graphs, and HPC, streamlining scientific exploration workflows.
- Built a multimodal RAG platform enabling search across 5+ billion industrial documents, including images and formulas, improving data access for engineers and operators.
- Fine-tuned Llama models with GRPO for domain-specific summarization, boosting insight extraction from maintenance and operational logs.
- Designed and launched an AI assistant for insurance policy analysis using LLMs, enhancing document understanding and reducing manual review time by over 60%.
Machine Learning Scientist
BASF
- Pioneered the application of deep learning to accelerate chemical and materials R&D, developing novel computational methods with direct parallels to challenges in computational biology and drug discovery.
- Engineered and optimized deep learning models (PyTorch, TensorFlow) for production environments, utilizing CUDA and NVIDIA Triton Inference Server for high-throughput, real-time scientific data processing.
- Designed and deployed a novel Variational Autoencoder (VAE) for multimodal latent space representation of complex scientific data, enabling advanced analytics and property prediction, leading to a patent (WO/2023/198927).
- Spearheaded the design of a digital quality assessment solution using CV and deep learning with an expert-in-the-loop, building an AI tool for high-throughput scientific analysis analogous to automated assay analysis (patent: WO/2023/285538).
Algorithm Development Engineer
ZF
- Collaborated in the autonomous driving department using Agile-Scrum.
- Developed a multi-sensor data visualization tool, enhancing AI-driven scene perception and data integration for obstacle detection, tracking, and sensor fusion.
- Developed a multi-sensor data visualization tool, enhancing AI-driven scene perception and data integration for obstacle detection, tracking, and sensor fusion.
Postdoc Machine Learning Researcher
Max Planck Institute
- Developed automated algorithms for analyzing advanced microscopy data using supervised/unsupervised ML.
- Designed and implemented a specialized CNN-RNN architecture, optimizing for processing large-scale scientific datasets.
- Led a research group (two PhD students), applying advanced mathematics and machine learning to complex scientific problems in materials science, including accelerating rare event simulations on HPC facilities.
- Contributed to AI and materials science research through publications in relevant peer-reviewed journals and conference presentations.
Experience
Autonomous Materials Discovery with Multi-agent LLMs
The system integrated literature and patent retrieval with dynamic knowledge graph construction and HPC simulations. It enabled collaborative agents to interpret scientific texts, propose new materials, and simulate outcomes, mirroring real-world R&D decision-making.
My role covered full-stack AI engineering, including fine-tuning LLMs, designing agent interactions, implementing vector search, and deploying the solution at scale.
The project significantly reduced time-to-discovery for materials scientists and set a new standard for intelligent, autonomous research tools.
Education
Postdoctorate Degree in Machine Learning
Max Planck Institute - Munich, Germany
Skills
Libraries/APIs
PyTorch, Hugging Face Transformers, Pandas, NumPy, Microsoft HPC, REST APIs, Keras, Scikit-learn
Tools
Azure Machine Learning, Azure OpenAI Service, Claude, StatsModels, NVIDIA Nsight Systems, Git, Azure ML Studio
Languages
Python
Frameworks
LlamaIndex, AutoGen, Hadoop
Platforms
Azure, Azure AI Search, Azure AI Studio, NVIDIA CUDA, Microsoft, Docker, NVIDIA NeMo, Google Cloud Platform (GCP)
Storage
Graph Databases
Industry Expertise
Teaching
Paradigms
Model Context Protocol (MCP), High-performance Computing (HPC)
Other
Deep Learning, Machine Learning, Physics, Azure AI Custom Vision, Retrieval-augmented Generation (RAG), Open-source LLMs, Large Language Model Operations (LLMOps), Large Language Models (LLMs), Generative Artificial Intelligence (GenAI), Multimodal GenAI, Agentic AI, Multimodal Models, Hugging Face, NVIDIA Triton, Knowledge Graphs, Multi-agent Systems, Artificial Intelligence (AI), AI Agents, LangChain, Natural Language Processing (NLP), RAG Architecture, Semantic Search, GRAPH, GraphDB, Data Science, Support Vector Machines (SVM), GPU Computing, CI/CD Pipelines, Data Visualization, Predictive Analytics, Architecture, Prompt Engineering, AI Development, Technical Instruction, Statistical Modeling, Spatial Analysis, ETL Pipelines, Azure AI Bot Service, GitHub Codespaces, Financial Markets, FastAPI, Software Architecture, Azure AI Document Intelligence, Neural Networks, Data Engineering, APIs, Google Cloud ML, Chemistry, Chemotherapy, Spectroscopy, Computer Vision, Data Scientist, Principal Component Analysis (PCA), Materials Science
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring