• As a machine learning engineer and data scientist in the search team at OTTO, my main task is to use state of the art machine learning techniques to improve the search experience for our customers.
• The Solr search engine, which processes 1.000 queries per second and supports around 20 million product variants 24/7, is central to OTTO's e-commerce platform.
• All improvements are extensively tested and validated through online experiments.

Learning to Select:

• Improved query precision by filtering out irrelevant results through comprehensive data-driven solutions on clickstream data. 
• Also identified and removed fraudulent and bot-generated queries to improve model performance and data integrity.

Hybrid Search:

• Collaborated with two teams to develop a system that integrates both lexical and
• semantic search approaches to provide more relevant search results.

Advanced Spell Check:

• Designed, implemented, validated and brought to production a leading-edge spell checking system. 
• This solution not only corrects customer spelling errors but also guides them towards the most relevant products.

Query Intent Detection:

• I also led the development of a customer query intent detection approach to identify non-product and navigation queries, and to recognize brand names and their context within search queries (Named entity recognition and classification).

• As an external consultant, I helped startups to use GPT and other large language models (LLMs).
• I provided training, evaluated use cases, assessed limitations such as security, performance, accuracy and explored options/alternatives to the OpenAI API.

Video based models:

• Aesthetic Ranking
• Dominant Color Extraction
• End Credits Detection
• Face Detection
• Image Quality Detection
• Logo Detection
• Mood Detection
• Object detection and Recognition
• Place Prediction
• Scene and Shot-Boundary Detection
• Shot Type Detection by using and optimizing pre-trained and self-trained models

Audio based models and solutions:

• Speech-to-Text transcriptions using Google?s Speech-to- Text API and Whisper from Open-AI on Podcasts and other audio sources

NLP solutions:

• language detection (fastText), festivity detection, kids content detection, adult content detection, topic modeling (BERTopic), keyword extraction (KeyBERT) and text summarization

• As a freelance consultant and expert in machine learning, data science and deep learning, I specialized in fraud detection, product recommendation systems, image recognition/classification, anomaly detection, time series analysis and NLP. I led agile projects from conception to production and maintenance & optimization.
• I focused on various e-commerce solutions that leveraged consumer data, product master data & descriptions, product images and sales transactions.

Product Similarity:

• in order to increase of the downstream system's performance this solution will help to find similar or related products for a particular product which can be used then as a benchmark or replacement
• The similarity will be determined by various modalities: 
  • visual similarity (image autoencoder), consumer behavior (clickstream data) and product descriptions (NLP transformers)

Skills:

Python, Jupyter, PySpark, TensorFlow, Jira, Bitbucket

Dynamic Pricing:

• the main goal for this project is to identify poor performing products in an early stage, uncover possible product issues and determine the right actions e.g. optimal price change to boost performance
• The overarching goal was to gradually replace the existing solution

Skills:

Python, Jupyter, PySpark, XGBoost, matplotlib, TensorFlow

Consumer Lifetime Value:

• conception, implementation and maintenance for the historical and future monetary value attributed to an individual consumer. Regular extensions and adaptations for e.g. new markets / brands and deep dive into the model's most important features
• The models are based on consumer behavior data and are running fully in production and will be updated on a weekly basis for all consumers. The results (KPIs) are intensively used in downstream systems and for marketing campaigns.

Skills:

Python, XGBoost, SHAP, matplotlib, Exasol, Jira, Bitbucket

Visual Product Embeddings:

• conception and implementation of a variational autoencoder based on product images
• The source images are being filtered, downscaled and prepared for a convolutional neural network (VAE) where the embeddings will be generated
• These embeddings are able to capture design elements of a product image which can be used to find similar products but also will be fed into downstream models to improve any productbased model
• The solution is running in production and will be updated with new images on a weekly basis

Skills:

Python, Keras/TensorFlow, PySpark, SageMaker, OpenCV

Purchase Propensity Scores: 

• conception, implementation and maintenance for modelling the consumer's purchase intention
• The solution is running very stable in production for a few years already and the results provide a high contribution to the marketing channels

Python, XGBoost, SHAP, matplotlib, Exasol, Jira, Bitbucket

• participating the "Histopathology Cancer Detection" competition
• The goal was to identify metastatic cancer in medical images
• My role was to bring state of the art computer vision techniques to the team and to implement an ensemble of models for the submission
• We have reached #26 from 1.149 competitors using advanced (high-speed) training techniques and heavy image augmentations

• The goal of the project was to build an MVP for a product image classification system to support annotators' workflows and to identify outliers/broken images within the image pool
• My role was also to educate the team on the latest deep learning/computer vision possibilities and find additional business cases to implement