Download CV Get in Touch
Available for opportunities

Hey, I'm KASHISH PATEL

AI / ML Engineer
Kashish Patel

About me

Kashish Patel

Hey 👋🏼 I am a Computer Science Master’s student based in Germany, specializing in Intelligent Systems with a passion for Artificial Intelligence and machine learning. My projects range from real-time face recognition to multi-view learning for Earth Observation and fine-tuning large language models for translation. With a background in Android development, I bring a user-focused approach to AI solutions, and I’m currently applying my skills through my internship at AICU and my thesis at DFKI, while exploring opportunities in Machine Learning, Multimodal AI, and applied AI research.

Experience

AUG 2025 — PRESENT
Multimodal AI Intern
AICU · Remote/Part-time
  • Researched and implemented fusion techniques for multimodal datasets.
  • Developed efficient training/evaluation pipelines; documented experiments for reproducibility.
  • Conducted efficiency comparisons across accuracy, training time, and inference speed.
JUL 2025 — PRESENT
Master Thesis Student
DFKI(German Research Center for AI) · Hybrid/Full-time
  • Conducting research on robustness in multi-view deep learning models for Earth Observation at DFKI.
  • Investigating the impact of input noise on latent space structure using probabilistic modeling techniques.
  • Developing uncertainty-aware fusion methods to enhance AI model interpretability and resilience in real-world scenarios.
  • Utilizing datasets such as CloudSen12, EuroSAT-SAR and Treesatai- TS to validate findings and improve model performance.
DEC 2021 — JUL 2022
Android Developer
Dominant Infotech — India
  • Collaborated with the Android development team to design, build, and test the SnackEatUp and DocsApp applications.
  • Developed Android apps using Java and Kotlin with MVVM architecture and Jetpack components.
  • Integrated APIs, managed local databases, and implemented user authentication.
  • Ensured app stability, usability, and performance through testing and debugging.
  • Implemented offline support with SQLite and Room Database.
  • Used Git for version control and participated in Agile sprints to meet release deadlines.

Education

OCT 2022 — PRESENT
M.Sc. Computer Science
RPTU Kaiserslautern-Landau · Major: Intelligent Systems · Minor: Software Engineering
  • Research Seminar: Survey of Multi-Modal Generative AI Tools in Education and Creative Industries.
  • Coursework: Deep Learning, Machine Learning, Social Web Mining, Data Visualization, Very Deep Learning, Collaborative Intelligence.
2018 — 2022
B.Tech Computer Engineering
CGPIT, Uka Tarsadia University
  • CGPA of 9.08/10

Projects

Deep Multi-View Information Bottleneck

  • Built and evaluated a multi-view framework combining satellite, radar, and weather data across LFMC, Yield Cereals, and PM2.5 datasets.
  • Benchmarked 7 fusion techniques, including stochastic VAEs and deterministic baselines, to analyze uncertainty modeling.
  • Achieved the best performance with feature-level stochastic fusion, reaching an average R² of 0.64 and RMSE of 23.4.
PyTorchMulti-ViewVAE/IBFusion
Code

LLM Fine-Tuning for German↔French Translation

  • Fine-tuned FLAN-T5 Large using LoRA on 800 real and 1,600 synthetic German–French sentence pairs from Europarl and Groq API.
  • Compared baseline, real-only, synthetic-only, and hybrid datasets to study fine-tuning and data augmentation impact.
  • Improved BLEU score to 20.8 (from 18.6) with hybrid training, showing gains in low-resource MT tasks.
PythonLoRA-FinetuningNLPGroq API
Code

Face Recognition Attendance System

  • Designed a Raspberry Pi + OpenCV–based real-time attendance system with >90% face recognition accuracy.
  • Integrated lightweight deep learning models for robust performance in variable lighting and classroom conditions.
  • Reduced frame processing time by 30% through NumPy optimization and asynchronous image capture.
OpenCVNumPyRaspberry Pi
Code

SnackEatUp – Food Ordering Android Application

  • Developed a full-stack Android application with category-based menus, cart management, and order placement.
  • Implemented Firebase for real-time authentication, database management, and order tracking.
  • Built as a team project to digitize snack ordering, improving service speed and reducing manual errors.
AndroidKotlinFirebase
Code

Skills

Languages

  • Java
  • Python
  • C/C++
  • SQL
  • HTML/CSS

Frameworks

  • TensorFlow
  • PyTorch
  • Hugging Face Transformers
  • scikit-learn
  • Keras
  • WordPress
  • FastAPI
  • LoRA Developer

Tools

  • Git
  • Google Colab
  • Raspberry Pi
  • Postman
  • Firebase Console
  • Android Studio
  • SQLite
  • Jupyter Notebook
  • Groq API

Libraries

  • NumPy
  • pandas
  • Matplotlib
  • Xarray
  • OpenCV

Get in Touch

Let's Talk

Have a role, project or research collab? I’m open.

Send a Message

Prefer email? kashishpatel017@gmail.com

© 2025 Kashishkumar Patel.