Projects

Innovations I’ve brought to life

Projects

Robotics and Autonomous Systems

Agriculture Autonomous Robot

  • Company/Institution: UnternehmerTUM, Munich, Germany
  • Description: Designed a robot to classify raspberry plant genders and spray unwanted species precisely.
  • Tools: OpenCV, PyTorch, CUDA, Jetson, TensorRT

Autonomous Robot with Navigation Module

  • Company/Institution: EPFL, Lausanne, Switzerland
  • Description: Developed a robot with navigation capabilities using vision and SLAM data.
  • Tools: OpenCV, PyTorch, ROS, Aseba

Guitar Playing Robot

  • Company/Institution: Technical University of Munich (TUM), Munich, Germany
  • Description: Built a robot that plays the guitar using MIDI signals and Arduino.
  • Tools: Arduino, C++

Computer Vision

Data Augmentation for Saliency Prediction

  • Company/Institution: EPFL, Lausanne, Switzerland
  • Description: Developed augmentation techniques to improve saliency prediction accuracy.
  • Tools: PyTorch, OpenCV, PIL

Visual Perception of Unknown Objects

  • Company/Institution: CARIAD (Volkswagen Group), Munich, Germany
  • Description: Developed instance segmentation and language-to-vision techniques, achieving state-of-the-art performance on COCO dataset. Submitted to CVPR.
  • Tools: Computer Vision, Transformers, Semantic Segmentation, Knowledge Distillation

Perception through Segmentation

  • Company/Institution: Chair of Visual Computing and AI, TUM, Munich, Germany
  • Description: Built PyTorch models for image segmentation and object detection.
  • Tools: PyTorch, Pandas

[License Plate Detection and Pedestrian Tracking]

  • Company/Institution: TUM, Munich, Germany
  • Description: Implemented OCR, GUI, and Kalman filters for license plates and pedestrian detection.
  • Tools: PyTorch, OpenCV, PyQt

Natural Language Processing (NLP)

From Judgement’s Premises Towards Key Points

  • Company/Institution: Technical University of Munich (TUM), Munich, Germany
  • Description: Extracted key points from premises in legal texts for automated analysis.
  • Tools: PyTorch, Bert, IBM Debater, PageRank, Clustering

Globywood: The Story of Movie Industries Across Time

  • Company/Institution: EPFL, Lausanne, Switzerland
  • Description: Analyzed cinema trends using advanced NLP methods.
  • Tools: NLTK, Bert, SentenceTransformer, SciPy

Edge AI and Model Optimization

Model Sparsification and Inference Optimization

  • Company/Institution: Intel, Munich, Germany
  • Description: Researched and implemented sparse, quantized neural networks for edge devices.
  • Tools: Python, PyTorch, TVM, SparseZoo, TFLite

Hardware Optimization for NFC Design

  • Company/Institution: Infineon Technologies, Munich, Germany
  • Description: Designed an NFC encoder/decoder to replace RAM in embedded systems.
  • Tools: VHDL, Verilog, Modelsim

Fleet of Robots Task Assignment

  • Company/Institution: Huawei Munich Research Center, Munich, Germany
  • Description: Optimized task assignment and monitoring for fleets of robots using reinforcement learning and LLMs.
  • Tools: PyTorch, TFLite, ROS, Quantization, Jetson, SLAM

Model Optimization for ECUs

  • Company/Institution: BMW, Munich, Germany
  • Description: Optimized neural networks for deployment on edge devices and designed a framework for model optimization.
  • Tools: PyTorch, TFLite, ONNX, Apache TVM

Hackathons and Uni Projects

Breast Cancer Detection and Localization from MRI Images

  • Company/Institution: Lauzhack, Lausanne, Switzerland
  • Description: Developed a CNN for detecting and segmenting breast cancer from MRI scans.
  • Tools: OpenCV, PyTorch, Docker, Transformers

Covid Interactive AI Dashboard

  • Company/Institution: Technical University of Munich (TUM), Munich, Germany
  • Description: Built an interactive AI dashboard for COVID-19 data visualization and prediction.
  • Tools: Python, Dash, Keras, Plotly

Multiplayer Chess with TCP Socket Server

  • Company/Institution: Technical University of Munich (TUM), Munich, Germany
  • Description: Developed a chess game with a TCP client-server architecture.
  • Tools: C++, Qt, TCP, OOP

Pre-Symptomatic Disease Detection Using Wearable Data

  • Company/Institution: TUM.AI, Munich, Germany
  • Description: Built an anomaly detection network for wearable device data to detect diseases early.
  • Tools: Python, PyTorch, Docker, NeuroKit2