Simon Benedict
B.Sc. Business Informatics — Machine Learning & OSS Enthusiast
Projects
Air Quality Forecast
Time series forecast for air quality in italian streets
Hallucination Detection
Comparison and evaluation of different methods to detect hallucinations in LLM outputs
LLM-based Chatbot
RAG (Retrieval-augmented Generation) chatbot leveraging Large Language Models to provide comprehensive information about Baldur's Gate 3
Anomaly Detection
Deep learning-based anomaly detection using CNNs to identify structural deviations and defects in LEGO Vespa models via image analysis
Skills
Python
- Solid experience in various contexts
- Web development (FastAPI, Flask)
- Deep learning (TensorFlow, PyTorch)
- LLM applications (LangChain, LlamaIndex)
I have significant experience with Python across various projects. I'm proficient in building web applications using FastAPI and Flask, and I've worked on machine learning projects using common frameworks like scikit-learn, PyTorch, TensforFlow and Transformers. Recently, I've been focusing on LLM applications, particularly RAG systems and agentic workflows, exploring prompt-chaining frameworks like LangChain, DSPy and AutoGen in my personal projects.
Data Analytics
- Data manipulation & analysis
- Visualization techniques
- Basic statistical analysis
- Experience with common data formats
I'm proficient in data manipulation and analysis using Pandas and NumPy. I have experience creating visualizations with Matplotlib and conducting basic statistical analyses. I've worked with various data formats and frequently use Jupyter notebooks for data exploration and analysis in my projects.
Machine Learning / AI
- Growing expertise in ML and Deep Learning
- Experience with common ML algorithms
- Work with deep learning models (CNNs, VAEs)
- Exploring LLMs and related frameworks
I've developed a strong foundation in Machine Learning and Deep Learning, primarily using Python. I've implemented various ML algorithms and built deep learning models like CNNs and VAEs. My recent focus has been on Large Language Models (and NLP more generally), working with APIs from OpenAI, Anthropic and similar, and hosting models locally using llama.cpp.
Java
- Foundation from university studies
- Understanding of OOP principles
- Applicable to software architecture concepts
I studied Java for three semesters at university, gaining a solid foundation in object-oriented programming. While I use Java less frequently now, the concepts I learned have been valuable in understanding software architecture across different languages and inform my approach to programming problems.
JavaScript / TypeScript
- Experience from academic and personal projects
- Familiar with core concepts and DOM manipulation
- Frontend development with React and Next.js
I've worked with JavaScript in both academic and personal projects, gaining familiarity with its core concepts and DOM manipulation. For frontend development, I use JavaScript and TypeScript with mostly with React and Next.js. I'm continuously expanding my skills in this area.
UI / UX
- Proficient in HTML and CSS
- Regular use in personal and academic projects
- Experience with TailwindCSS
I have a strong grasp of HTML and CSS, which I use regularly in personal and academic projects. I've also gained experience with TailwindCSS for creating responsive designs.
Databases
- Experience with relational and unstructured databases
- Proficient in SQL, PHP, and Java for databases
- Work with vector DBs (ChromaDB, Pinecone) for NLP
My studies provided a solid foundation in database concepts and design, working with SQL (MySQL & SQLite), PHP, and Java for relational databases. For NLP projects, I've explored vector databases like ChromaDB and Pinecone.
Cloud
- Experience with Google Cloud Platform (GCP)
- Conceptual knowledge from bachelor's studies
I have practical experience with Google Cloud Platform from using it in my personal projects. Additionally, my bachelor's studies provided me with a theoretical understanding of cloud concepts.
Let's build
You can contact me via
[email protected]
or use the form below: