Summary
Overview
Work History
Education
Skills
Projects
Timeline
Hi, I’m

Zahid Pathan

Munich
Zahid Pathan

Summary

Experienced in retrieval AI techniques through an internship at BMW Group, currently working on a thesis in solution spaces for vehicle design using AI/ML and training deep learning models. Additionally skilled in software development, advanced FEM, and 3D modeling.

Overview

1
year of professional experience
6
years of post-secondary education
2
Languages

Work History

BMW Group

Master Thesis
11.2024 - Current

Job overview

  • Developing a methodology to implement solution spaces, an important concept used in early phases of vehicle design, by utilizing AI regression models, replacing traditional approaches
  • Implementing a convolutional encoder-decoder network to predict the B-pillar acceleration time series in a full-frontal crash test

BMW Group

Intern
04.2024 - 09.2024

Job overview

  • Developed an AI-driven Retrieval-Augmented Generation (RAG) chatbot with a user-friendly interface, utilizing large language models (LLMs) and CUDA-accelerated vector embeddings to deliver accurate responses for user queries on over 200 BMW internal modeling guideline documents
  • Benchmarked retrieval techniques, evaluating various architectures and configurations to achieve a 90% accuracy rate, significantly improving response reliability and user experience
  • Collaborated with service partners to optimize the crash management system model in ANSA, incorporating a semiautomated workflow to boost efficiency. Applied machine learning techniques to analyze vehicle crash components

Atlas Copco

Working Student
06.2023 - 02.2024

Job overview

  • Analyzed and extended results of an ICP algorithm to assist bolt position detection, by extracting data from point clouds
  • Classified the position and orientation of a 3D scanned bolt with an precision threshold of 0.5mm, using inlier ratio as the key metric for determining correct alignment
  • Designed libraries to integrate a TOF camera into the software and trained deep neural network (DNNs) models for object detection and positioning to assist manufacturing processes

Education

Technical University Munich (TUM)
Munich, Germany

Master of Science from Computational Mechanics
10.2022 - Current

University Overview

  • Specializations: AI in CoMe, Advanced FEM, Car Body Design and Safety Failure Mechanics
  • 1.8 GPA

Indian Institute of Technology (IIT)
Dharwad, Karnataka

Bachelor of Technology from Mechanical Engineering
08.2018 - 07.2022

University Overview

  • Silver Medalist - for achieving the highest GPA
  • GPA: 1.3

Skills

Programming Languages & Frameworks: Python, C/C, PyTorch, TensorFlow, Keras, LangChain, Gradio, NLP, Training and Fine-tuning LLMs, Git, HuggingFace, OpenCV, OpenMP, Docker, Matlab, CMake, Jira

Projects

Projects

Prediction of additive manufacturing process using DNNs

  • Developed a feed-forward neural network using TensorFlow and Keras to predict layer width and failure points in 3D printed clay, leveraging deep learning techniques and achieving an 82% accuracy rate
  • Implemented an end-to-end MLOps pipeline to streamline data preprocessing, hyperparameter tuning, and model optimization, aligning with state-of-the-art techniques for robust iteration
  • Utilized custom loss functions and created evaluation frameworks to monitor prediction accuracy, enabling quality improvements, which reduced material waste and enhanced print stability

Timeline

Master Thesis
BMW Group
11.2024 - Current
Intern
BMW Group
04.2024 - 09.2024
Working Student
Atlas Copco
06.2023 - 02.2024
Technical University Munich (TUM)
Master of Science from Computational Mechanics
10.2022 - Current
Indian Institute of Technology (IIT)
Bachelor of Technology from Mechanical Engineering
08.2018 - 07.2022
Zahid Pathan