Alexander H.Private profile: Upwork shows its earnings only to signed-in members, so we check it every few weeks, not daily. MRR here is a rough estimate (Upwork's six-month earnings ÷ months active). Last read Nov 5, 2026.

⏩ More than 5y experience in Python, ML and data science in academia and industry, coupled with a commitment to staying up-to-date ⏩ Convinced that comprehensive communication is key to ideally complete your project, just as delivering clean, well-documented code ensures long-term maintainability ⏩ PhD in machine learning for cybersecurity, so I can build your ML stack and ensure its secure deployment I love working on challenging tasks. My broad interest in ML techniques allows me to pick the right method for any task and my mathematical understanding allows me to put ideas from cutting-edge research into practice. From building your data lake, over designing a custom neural network that suits your data best, to ensuring its scalable Cloud-based or on-prem deployment, I am well-prepared to delivering impactful results. 𝗙𝗿𝗲𝗾𝘂𝗲𝗻𝘁𝗹𝘆 𝗨𝘀𝗲𝗱 𝗧𝗼𝗼𝗹𝘀 • Python, C++ • Keras, Jax, tensorflow, PyTorch, HF Transformers for 🧠 Deep Learning • AWS Sagemaker AI, Firehose, Athena, S3, Lambdas for ☁️ Cloud-based data (pre)processing and model deployment • scikit-learn, pandas, SciPy, XGBoost, pyod, matplotlib, seaborn for 📊 data mining & visualization • librosa, pyaudio, pyroomacoustics, madmom for 🎶 audio ML tasks • OpenCV, PaddleOCR, YOLO for 🎨 CV tasks • LlamaIndex, LangChain for RAG • GitHub Actions, Docker containerization, Amazon ECR for MLOps • FastAPI, flask, nginx, MySQL/Postgres web backend stack • GCE, AWS EC2 and CUDO Compute for 🖥️ Cloud VMs/GPUs • Terraform for IaC 𝗦𝗲𝗹𝗲𝗰𝘁𝗲𝗱 𝗣𝗿𝗼𝗷𝗲𝗰𝘁𝘀 👉 𝘐𝘯 𝘾𝙤𝙢𝙥𝙪𝙩𝙚𝙧 𝙑𝙞𝙨𝙞𝙤𝙣, 𝘐 𝘪𝘮𝘱𝘭𝘦𝘮𝘦𝘯𝘵𝘦𝘥 • a full OCR pipeline with subsequent PDF data extraction, based on the Differentiable Binarization model for text detection, the PARSeq ViT model for text recognition and transformers using multidimensional RoPEs • image segmentation with BASNet. • object detection, classification and segmentation based on the Mask RCNN architecture. 👉 𝘐𝘯 𝘵𝘩𝘦 𝘤𝘰𝘯𝘵𝘦𝘹𝘵 𝘰𝘧 𝙩𝙧𝙖𝙣𝙨𝙛𝙤𝙧𝙢𝙚𝙧𝙨, I engineered highly optimized custom layer implementations in C++/CUDA, including both forward pass and manual gradient backpropagation. 👉 𝘍𝘰𝘳 𝙡𝙖𝙧𝙜𝙚-𝙨𝙘𝙖𝙡𝙚 𝙙𝙖𝙩𝙖 𝙚𝙣𝙜𝙞𝙣𝙚𝙚𝙧𝙞𝙣𝙜, 𝘐 𝘥𝘦𝘷𝘦𝘭𝘰𝘱𝘦𝘥 a scalable data lake solution based on Terraform-deployed AWS services, leveraging Amazon Firehose, Amazon S3 and AWS Lambdas for data ingestion, Amazon EventBridge for real-time event propagation and Amazon Athena and FastAPI for query and endpoint development. 👉 𝘐𝘯 𝙘𝙮𝙗𝙚𝙧𝙨𝙚𝙘𝙪𝙧𝙞𝙩𝙮, I conducted comprehensive data analysis, algorithm development and ML model design in the context of network traffic analysis during my PhD journey. 👉 𝘐𝘯 𝙖𝙪𝙙𝙞𝙤 𝙈𝙇, 𝘐 • led research on generative audio ML, testing various architectures. • performed MIR model implementation and improvements for a major Austrian startup. 👉 𝘐𝘯 𝘴𝘮𝘢𝘭𝘭𝘦𝘳 𝘱𝘳𝘰𝘫𝘦𝘤𝘵𝘴, 𝘐 𝘢𝘭𝘴𝘰 𝘤𝘰𝘯𝘵𝘳𝘪𝘣𝘶𝘵𝘦𝘥 𝘵𝘰, 𝘦.𝘨., • VAEs in the context of molecule discovery. • GANs for image generation. 𝗠𝘆 𝗕𝗮𝗰𝗸𝗴𝗿𝗼𝘂𝗻𝗱 • Education in electrical engineering, which gives me a strong theoretical underpinning • Extensive writing experience as evidenced by more than 10 academic papers, several authored grant applications and a doctoral thesis nominated for the Austrian national Zemanek prize • Used various deep learning techniques including RNNs, CNNs, MLPs, transformers throughout previous work • Deployed various ML and data mining techniques including 𝘬NN, XGBoost, Random Forests, clustering and anomaly/outlier detection in previous projects • Azure Cloud Certification 𝘓𝘦𝘵'𝘴 𝘣𝘶𝘪𝘭𝘥 𝘴𝘰𝘮𝘦𝘵𝘩𝘪𝘯𝘨 𝘨𝘳𝘦𝘢𝘵 𝘵𝘰𝘨𝘦𝘵𝘩𝘦𝘳.