Valentin Lukas

Über mich

Ich unterstütze Unternehmen dabei, sich von AI-Prototypen zu sicheren Produktionssystemen auf Enterprise-Niveau weiterzuentwickeln, und minimiere dabei die Risiken bei der Implementierung unter dem EU AI Act. Unterstützt durch die Sicherheits-Expertise von KASTEL agiere ich als Technical Lead oder Architekt, um Data-Science-Teams auf ein neues Level zu bringen und robuste MLOps-Frameworks innerhalb einer regulierten Enterprise-IT zu implementieren. Kernfokus: -🚀 MLOps & Architektur: Aufbau skalierbarer Pipelines auf Azure & Databricks. - 🛡️ KI-Sicherheit: Adversarial Machine Learning & Compliance (EU AI Act). - 💡 Technische Führung: Überbrückung der Lücke zwischen F&E-Experimenten und dem IT-Betrieb. Aktuell verfügbar für strategische Beratung & Architekturrollen im Bereich Data & AI (Remote/Hybrid).

Skills

Expert:in

AI Consulting

AI Development

Python

PL/SQL

AI Application

Analytical Thinking

Data Analysis

Data Science

Deep Learning

Künstliche Intelligenz

Machine Learning

Neuronale Netze

Reinforcement Learning

Self-learning Systems

Technische Dokumentation

Fortgeschritten

AI Ethics and Governance

Azure Cloud

Generative AI

Data Governance

Data Engineering

Data Strategy

Agile Development

Data Visualisation

CI/CD

Cloud Computing

Continuous Deployment

Computer Vision

Consumer Intelligence Analysis

Data Design

Docker

DevOps

Datenschutz

FastAPI

Github

GitLab

Git

Kanban

Kubernetes

MySQL

Pattern Recognition Systems

PostgreSQL

Prompt Engineering

Prometheus

REST-API

Scrum

Softwareentwicklung

Software Engineering

Software Design

Solution Architecture

Visual Studio Code

Wissenschaftliches Schreiben

Grundkenntnisse

BigQuery

ChatGPT

Claude

Continuous Integration

Copilot

Cyber Security

Data Migration

Elasticsearch

Excel

Google Cloud Platform

Kryptografie

Stakeholder Management

Storytelling

Projekte

CX Analytics @ Major Telco Provider | Role: Lead Data Scientist & MLOps Architect
· Internet und Informationstechnologie · 1.000+ Mitarbeiter:innen
Januar 2024 — heute
Stream A: CX Analytics & NPS Optimization (Current focus):

Spearheading the end-to-end development of causal ML models (Uplift Modeling) to improve customer experience.

Establishing MLOps standards using CI/CD, GCP, Oracle and Databricks for automated cloud infrastructure.

Development of "Explainable AI" components (SHAP) to translate complex model decisions for business stakeholders.
Financial Risk @ Major Telco Provider | Role: Lead Data Scientist & MLOps Architect
· Internet und Informationstechnologie · 1.000+ Mitarbeiter:innen
Januar 2024 — heute
Stream B: Realtime Financial Risk Application:

Architected the migration of critical risk application to a scalable Azure/Databricks environment.

Model optimization and software development for team MLOps strategy.

Compliance & Governance: Ensured strict adherence to GDPR (BDSG), Banking Secrecy, and TKG regulations through audit-proof data pipelines.

Innovation:

Conducted PoCs for Vector-Database integration (CosmosDB) to modernize risk scoring

Design of custom documentation structure for component-rich architecture (Confluence)

Development of a standard for exploratory data analysis

Environment: agile Scrum, Python, scikit-learn/sklearn, MLflow, Tensorflow, Keras, Jira, incl. monitoring, alerting, Apache Spark, ACA, Docker, Kubernetes, KQL, stakeholder engagement, Databricks, Unity Catalog.
HCP Targeting Platform | Role: Technical Product Lead
· Gesundheit und Soziales · bis zu 10 Mitarbeiter:innen
November 2025 — Februar 2026
Product Vision: Developed a data platform for the segmentation of Healthcare Professionals (HCPs) to optimize pharmaceutical targeting strategies.

Tech Stack: Built a robust Python package with automated data pipelines, analysis and visualization.

Advanced Analytics: Implemented community detection algorithms and similarity clustering (Jaccard) to map HCP networks and influence clusters.

Quality Assurance: Established a full test suite and automated data quality checks.

Environment: Python, Parquet, Pandas, NetworkX, PyTest, Agentic Workflow Software Development

Alle Projekte ansehen

Berufserfahrungen

CEO and Data Scientist · Vollzeit
· Internet und Informationstechnologie · 1 Mitarbeiter:innen
Januar 2024 — heute
Project: CX Analytics and Financial Risk @ Major Telco Provider | Jan 2024 – Present | Role: Lead Data Scientist & MLOps Architect

Stream A: CX Analytics & NPS Optimization (Current focus):

Spearheading the end-to-end development of causal ML models (Uplift Modeling) to improve customer experience.

Establishing MLOps standards using CI/CD, GCP, Oracle and Databricks for automated cloud infrastructure.

Development of "Explainable AI" components (SHAP) to translate complex model decisions for business stakeholders.

Stream B: Realtime Financial Risk Application:

Architected the migration of critical risk application to a scalable Azure/Databricks environment.

Model optimization and software development for team MLOps strategy.

Compliance & Governance: Ensured strict adherence to GDPR (BDSG), Banking Secrecy, and TKG regulations through audit-proof data pipelines.

Innovation:

Conducted PoCs for Vector-Database integration (CosmosDB) to modernize risk scoring

Design of custom documentation structure for component-rich architecture (Confluence)

Development of a standard for exploratory data analysis

Environment: agile Scrum, Python, scikit-learn/sklearn, MLflow, Tensorflow, Keras, Jira, incl. monitoring, alerting, Apache Spark, ACA, Docker, Kubernetes, KQL, stakeholder engagement, Databricks, Unity Catalog.

Project: Parts Risk Scoring System (Supply Chain Optimization) | Nov 2025 – Jan 2026 | Role: Data Strategist & Senior Data Scientist

Business Impact: Designed a multi-dimensional scoring engine to identify "problem parts" in a complex supply chain, directly addressing stockouts and revenue gaps.

Architecture: Built a modular, explainable scoring system on an ELK Stack (Elasticsearch, Kibana) to process disparate data sources (ERP, inventory, supplier performance).

Leadership: Advising stakeholders on data strategy and ownership structure.

Environment: Dagster, Python, Polars, ElasticSearch, Agentic Workflow Software Development

Project: HCP Targeting Platform | Nov 2025 – Feb 2026 | Role: Technical Product Lead

Product Vision: Developed a data platform for the segmentation of Healthcare Professionals (HCPs) to optimize pharmaceutical targeting strategies.

Tech Stack: Built a robust Python package with automated data pipelines, analysis and visualization.

Advanced Analytics: Implemented community detection algorithms and similarity clustering (Jaccard) to map HCP networks and influence clusters.

Quality Assurance: Established a full test suite and automated data quality checks.

Environment: Python, Parquet, Pandas, NetworkX, PyTest, Agentic Workflow Software Development
AI Expert · Vollzeit
· Internet und Informationstechnologie · 10–25 Mitarbeiter:innen
Oktober 2020 — Dezember 2023
Project: Realtime AI Financial Risk Application Migration & Development (Telco) | Role: Lead Data Scientist

Big Data Engineering: Developed ETL pipelines handling massive datasets, event modelling.

Development of new functionalities (analytics modules, user feedback and model training processes) for internal analytics platform.

Development of SQL and PL/SQL load runs based on specific requirements and storage of the results in relational SQL databases.

Anomaly Detection: Built models for time-series anomaly detection and automated root cause analysis, including explainable AI (SHAP)

Operational Excellence: Created monitoring dashboards to ensure 24/7 system reliability.

Environment: SQL and PL/SQL (Oracle), Apache Spark, Hadoop, Apache Hive, Python and Kubernetes/Docker, Azure, AWS; Keras, scikit-learn/sklearn, MLflow; Jira and Confluence, Grafana, Prometheus, Tableau, Databricks, Unity Catalog.

Project: Speech Recognition Bot (Tech Startup) | Role: Data Scientist

Implemented automatic speech recognition (Kaldi, Shell) and Named Entity Recognition (NER) for a conversational AI interface.

Containerized software operationalization.

Managed stakeholder expectations and defined the technical roadmap.

Project: Address Classification (Logistics) | Role: Assisting Data Scientist

Implemented ETL pipeline using Kedro + PyTorch

Inference API operational via FastAPI and Docker

Data analysis using Matplotlib, Jupyter Notebooks, Python, on Azure

System stress testing with locust

Operationalizing LLM

Alle Berufserfahrungen ansehen

Professionals have standards

Über mich

Skills

Projekte

Berufserfahrungen

So funktioniert Junico

Valentin Lukas ist Freelancer:in bei Junico