Heidelberg University 1.3 · Very Good
Oct 2023 – Jan 2026
M.Sc. Data and Computer Science · Heidelberg, Germany
Y
Yusuf Berber
Data Science & Software Engineering · M.Sc. Heidelberg University
yusufberber19@gmail.com
Summary
Data science and software engineering professional completing an M.Sc. in Data and Computer Science at Heidelberg University (Grade: 1.3). Holds a B.Sc. in Software Engineering from Heilbronn University (Grade: 1.4). Published an official open-source LangChain integration at SAP SE. Master's thesis achieved State-of-the-Art results on benchmark emotion recognition datasets by combining fine-tuned LLMs with RAG. Fluent in Turkish (native), German (C2), and English (C1/C2).
Education
Heilbronn University 1.4 · Very Good
Mar 2019 – Aug 2023
B.Sc. Software Engineering · Heilbronn, Germany
Skills
Programming
ML & AI
Infrastructure
Languages
Experience
SAP SE
Working Student – Software Engineering · Walldorf, Germany
Nov 2024 – May 2025
Built and released the official LangChain integration package langchain-hana for SAP HANA Cloud, enabling vector search and knowledge-graph-backed retrieval in LLM applications. Implemented keyword search, in-database embeddings, and graph-based retrieval with full unit test coverage. Automated CI/CD quality gates with GitHub Actions and contributed upstream feature parity to LangChain (Python) and LangChainJS (TypeScript).
↗ github.com/SAP/langchain-integration-for-sap-hana-cloud
Vector Informatik GmbH
Intern – Software Engineering · Stuttgart, Germany
Mar 2021 – Sep 2021
Working Student – Software Engineering · Stuttgart, Germany
Apr 2022 – Aug 2022
Contributed to the ELEKTRA process tool within a high-performance automotive software platform. Implemented server-side reports and validators in Java, developed client-side features in C#, and maintained GUI tests with Ranorex.
Selected Projects
Emotion Detection in Conversational AI
2025 · Master's Thesis
Fine-tuned LLaMA-3.1-8B-Instruct via two-phase QLoRA and combined engineered audio features with retrieval-augmented in-context exemplars (RAG). Achieved State-of-the-Art performance on MELD and IEMOCAP benchmark datasets.
↗ github.com/yberber/mm-rag-erc
RAGMedAssist – Medical Chatbot
2023 – 2024
Co-developed an end-to-end LLM-powered medical assistant with retrieval-augmented generation pipelines tailored to the medical domain.
↗ github.com/Matteo-Malve/RAGMedAssist
Anime Face Generator Pipeline
2024
Built and trained a GAN-based image generation pipeline using DCGAN and WGAN-GP with PyTorch, combined with Real-ESRGAN for high-quality upscaling.
↗ github.com/yberber/anime-face-gan-pipeline
AlphaZero for International Checkers
2023 · Bachelor's Thesis
Re-engineered the AlphaZero algorithm using Monte Carlo Tree Search (MCTS) and ResNets. Built a simulation environment for synthetic training data via self-play.
↗ github.com/yberber/Checkers-AI-Minimax-Neural_Network