NumPy Experts in Germany

Project: Defect Generation on Test-Bench Images of Metal Surfaces Environment: Automated Visual Inspection (AVI), Metallurgy & Manufacturing

• Objective & Implementation: Designed, architected, and trained Generative Adversarial Networks (Pix2PixHD / SPADE) for image-to-image transformation. Targeted generation of synthetic material defects (e.g., cracks, inclusions, scale) on rough metal surfaces under real test-bench lighting conditions for privacy-compliant and efficient dataset expansion (data augmentation).
• Technical Design: Implemented robust Generative AI and computer vision pipelines in Python and PyTorch. Used semantic segmentation approaches for mask-controlled defect synthesis and subsequent evaluation with EfficientDet object detection models.
• Business Impact: Massive dataset upscaling (10x) without time-consuming and costly physical test-bench runs, while significantly improving the detection performance of automated inspection systems.

Technologies & Skills Used: Python | PyTorch | SPADE | Pix2PixHD | EfficientDet | Machine Learning | Semantic Segmentation | Computer Vision

• Development of an AI-powered project organization tool for a civil engineering company that intelligently links project, task, tender, schedule, and document data through a knowledge graph.
• Implementation of AI features for document analysis, information extraction, context-based assistance, and voice-based data capture based on Microsoft Azure AI, reducing administrative effort, making information available faster, and supporting project teams in decision-making.
• Tech stack: Python, React, TypeScript, FastAPI, Claude Code, Codex, Graphify, PostgreSQL, Microsoft Azure AI Foundry, Azure OpenAI, Azure AI Speech, Azure AI Document Intelligence, Microsoft Graph, Microsoft Entra ID, Docker, Git, CI/CD.

• This project involved extraction of intelligence data to support report writing for a company that provides geopolitical, global, commercial intelligence. The data have been gathered from a number of resources (interview transcripts, online data, internal documents), and then a knowledge base has been build from it. This was the basis of a complex RAG system, that was evaluated against a golden dataset. Agents have been used to find out the contradicting intelligence, the statements supporting each other, and to store back the generated knowledge.

Used: Python, RAG, LangGraph, LangChain, deepeval, MCP

• Task Engineering: Evaluating Coding Agents.
• Asset Creation: Building realistic task environments using datasets and files in German. Crucially, these assets must remain in the target language to genuinely measure multilingual handling.
• Prompting & Translation: finding failure points where AI does not work, in German.
• Implementation & Verification: Supporting the development of robust solutions (reference implementations) and write highly reliable, deterministic verifier scripts (using rubric-based judging only when strictly necessary).
• Calibration & Execution: Analyze execution logs and calibrate task difficulty (Easy to Very Hard) using standard Terminal-Bench run configurations against various model tiers (Haiku, Opus).
• Quality Assurance: Participation in a rigorous, 4-layer human quality control process (creation, human review, calibration review, and audit) alongside automated LLM-based checks to ensure fairness, grammatical accuracy, and benchmark integrity.
• Linguistic Review: Reviewing AI benchmark tasks across Hindi, Arabic, Japanese, Chinese, Czech and Turkish.

Most mid-sized companies already have their numbers. What is missing is the translation: a dashboard with forty tiles does not answer a single question that is actually asked in management.

Analysis

• Evaluation of existing data with Python and SQL
• Checking data quality and methodology before making a statement
• The result is an analysis that leads toward a concrete decision

Preparation

• Reports in Power BI and Tableau
• Interactive calculators and visualizations on the web
• Presentations and specialist texts for customers, sales and the public
• Analysis and communication from one source — that

• Analysis of the existing CAN communication between microcontrollers
• Analysis of the sensors used and the measured values collected
• Planning the CAN messages for transmitting the measured values
• Iterative adjustment of the microcontroller code to the new CAN messages
• Cross-compilation from x64 to arm64

• Developing a model-based AI where natural language is interpreted as formal relations.
• Natural language terms are not considered rigid but fluid and can be negotiated in a context so meaning resolves by iteratively specifying.
• Develops all kinds of model solutions.
• Backed by natural language and data annotation.
• Requirements to code and other solutions.

• Building, implementing and selling automated Data Science solutions such as Scorecard Factory and Forecast Factory
• Implementation of automated end-to-end cloud processes
• Development of LLM and NLP models
• Creation of interactive reports
• Support for national and international large corporations as well as medium-sized companies in implementing ML projects

Development and further development of an internal platform for managing and providing technical resources, virtual machines, and infrastructure services. The platform supports self-service processes and covers functions that are conceptually comparable to cloud management solutions like Azure or AWS.

• Responsible involvement in the design, development, and implementation of new backend and frontend features
• Hands-on development with Java, Spring Boot, Python, and Angular
• Implementation of REST interfaces, business logic, validations, and integrations into existing system landscapes
• Further development of modern web interfaces with Angular, including connection to backend services
• Participation in architecture and design decisions within the team, especially with regard to scalability, maintainability, and clean interfaces
• Containerization and deployment of applications with Docker, Kubernetes, and Helm
• Support with CI/CD processes and deployment to Kubernetes-based environments
• Work in the environment of vSphere, Broadcom, GitLab CI/CD, ArgoCD, Maven, npm, and NuGet
• Close collaboration with developers, business teams, DevOps, and other technical stakeholders
• Analysis of technical requirements, deriving suitable solutions, and independent implementation in an agile team
• Use of GitHub Copilot to support code generation, refactoring, test case creation, and technical documentation

Methods/ tools/ technologies: Languages & frameworks: Java (21), Spring Boot (4.x), Python, Angular, Robot Framework, Kubernetes, Helm Persistence: PostgreSQL, MongoDB, Hibernate, Liquibase Architecture & communication: REST, gRPC, GraphQL, Apache Kafka, OpenAPI, Microservices, Event Driven, Domain Driven Design Cloud & infrastructure: Terraform, Docker, Rancher, Helm, Ansible Security: OAuth2, MS (Entra ID), web security, Keycloak (extensions for detailed group rights) DevOps: GitLab CI/CD, Ansible, Maven, Gradle, Grafana, Prometheus, Git, GitHub Copilot Testing & QM: JUnit, Robot Framework, automated component and integration tests, E2E tests with Playwright, Testcontainers, EasyMock Methodology & approach: Kanban, JIRA, Confluence, Clean Code

Responsibility for commercial finance in Central Europe (DACH and Poland), reporting to the EMEA Commercial Finance Director. Monthly financial reporting, forecasting, and variance analysis, evaluation of promotions and special campaigns, management of planning processes including budgeting, as well as preparation of QBR materials up to CFO level. Took over functional leadership in the finance team after the mandate holder was unavailable.

• Built and fine-tuned CNN and RNN architectures using transfer learning for real-world classification tasks — core deep learning skills applicable to BMW's multimodal LLM and GenAI vehicle function development.
• Implemented Dropout, Batch Normalisation, and Early Stopping across deep learning experiments; evaluated rigorously using precision, recall, F1-score, and confusion matrices for production-grade reliability.
• Developed an AI-powered attendance management system using LBPH facial recognition, deployed via Flask web interface with real-time SMS notifications — demonstrating end-to-end AI product delivery for real users.
• Collaborated across cross-functional teams to deliver scalable, documented ML pipelines designed for reproducibility — matching BMW's interdisciplinary team and research environment.
• Integrated AI tooling directly into the development workflow from design through to testing, maintaining high velocity without compromising correctness.

• Title: Enhancing Syntactic Awareness in Transformer Language Models for Hindi Dependency Parsing
• Investigated syntactic knowledge captured by transformer language models (RoBERTa, XLM-RoBERTa) for Hindi dependency parsing, a morphologically rich and low-resource language.
• Developed structure-aware model variants (Struct_Roberta_hi, Struct_XLMR) by integrating a CNN-based parser network between transformer layers, inspired by the StructFormer architecture.
• Conducted extensive error analysis including label-wise, distance-based, direction-based, sentence length-based, and LVC/Non-LVC evaluations.
• Evaluated models on downstream NLP tasks (NER, POS tagging) using the IndicXTREME benchmark.

• Architected and deployed end-to-end machine learning pipelines across classification and prediction datasets, ensuring robustness and reproducibility through MLOps best practices.
• Contributed directly to LLM model output accuracy improvement by designing and engineering specialised prompts grounded in end-to-end ML and SciML pipeline logic.
• Developed training data for large language models by formulating coding problems that models could not resolve and subsequently documenting the correct solutions.

• Led strategic software development of fusion algorithms for precise object tracking, trajectory prediction, and environment modeling based on multimodal sensor data (e.g., camera, LiDAR, radar, GNSS, IMU)
• Developed and implemented navigation algorithms for autonomous vehicles, including path planning, obstacle avoidance, and sensor fusion of visual, inertial, and distance-based sensor sources
• Automated extraction and training processes with CI/CD
• Developed and optimized data pipelines and processes in Microsoft Azure using Apache Spark, Databricks, and PySpark
• Developed and optimized embedded software for automotive control units
• Designed latency-critical software for real-time control in robotic systems with RTOS (freeRTOS, SAFERTOS)
• Used the Vector toolchain (CANdela, DaVinci, CANoe) for configuration and diagnostics
• Optimized existing data pipelines and processes (ETL, data warehouse, SQL)
• Developed and trained machine learning models using PyTorch
• Created deep-learning-based object detection and visual SLAM algorithms, trained on combined data from camera, LiDAR, and IMU sensors
• Implemented computer vision algorithms for object detection and classification in robotic systems using OpenCV and YOLO, utilizing synchronized image and depth data
• Implemented behavior-based control systems for autonomous robots using ROS2 Behavior Trees
• Performed testing, release, and integration of sensor fusion algorithms into automotive production programs
• Ensured adherence to proper software development processes and safety standards to guarantee high data quality (MISRA, ISO 26262, ASPICE)