Implement a command line?based tax automation and pdf processing tool

Fine?tune Llama2 LLM with custom generated training data to enable custom information retrieval from pdf documents

Utilize student?teacher approach to generate training data with ChatGPT results

Conceptualize functional & non?functional requirements, consult business requirements and technical specification teams, technical review of specifications

Develop technical design for real?time streaming application processing multi?billion messages per day

Implement multiple Spark Structured Streaming applications, including custom outer?join operations, Hbase access and complex data structures

Conceptualize and implement data migration routines from classical DWH to event stream?based representation

Help client with stability and performance challenges with a large?scale single node Neo4j database

Implement a concurrent ETL process to load a timely?sorted two?dimensional grid of linked nodes in Neo4j

Setup a monitoring system for Neo4j

Implement an automatic backup process to backup the Neo4j database nightly and store archives

Design, implementation & operation of a large?scale, stream?fed, multi?billion node knowledge graph (TigerGraph)

Requirement analysis, design, implementation and operation of a graph?based real?time recommendation engine for news articles

Schema modeling & validation of a knowledge graph of multiple data streams

Design and implement real?time streaming applications

Give internal workshop about Apache Kafka

Designed and implemented a proof?of?concept for an event?driven data analytics application

Developed an ingestion pipeline to transform and feed relational data into a graph database

Optimized a large?scale text analytics pipeline for scalability and performance

Implemented knowledge discovery use?cases utilizing state?of?the?art NLP & ML approaches

Researched and applied efficient algorithms for analyzing large data sets

Educated in?house development team on big data software development and data mining

Project: Automatic calculation of nutritional values for food recipes

Consultation on planning of the project and the solution approach

Solution design: POC for a calculation pipeline based on multiple heterogeneous data sources and various machine learning approaches

Implementation of data processing pipeline to calculate nutritional values

Deploy pipeline on client infrastructure (AWS)

Title: Knowledge Discovery in textual Databases for enhancing the

automatic Calculation of nutritional Values for online?based Food Recipes

The work utilizes various machine learning and NLP approaches to extract information from unstructured text to determine the nutritional content of food recipes

Theoretical approaches: Tokenization, Part?of?Speech Tagging, Stemming, Neural Networks, Logistic Regression, Word Embeddings

As a team of interdisciplinary TUM students we developed a child?sized humanoid robot and promote the work on various events all over the world

As the team leader of the software development group, I was responsible for design, implementation and software engineering processes of the group

Project: Provide an indicator for the healthiness of food recipes which is based in expert ratings

Consulting in management and design thinking to derive useful innovative use cases from existing data of the client

Proof of concept: supervise generation of training data set and analyze data quality

Implementation of supervised classification system based on selected features and various approaches

Implementation and validation of final solution

I was involved in the software development team developing a time? and safety?critical, distributed middleware in C

The middleware was deployed on a highly modular avionics platform developed by ESG

The platform was composed of multiple Unix?like modules, which were interchangeable on the fly