I am Ashish Desai, a passionate AI and data science. I would like to make the AI solutions for the challenging real-time applications by playing with complex-unstructured data.

Ashish Desai

I am Ashish Desai, a passionate AI and data science. I would like to make the AI solutions for the challenging real-time applications by playing with complex-unstructured data.

Available to hire

I am Ashish Desai, a passionate AI and data science. I would like to make the AI solutions for the challenging real-time applications by playing with complex-unstructured data.

Experience Level

Expert
Expert
Expert
Expert
Intermediate
Intermediate
Intermediate
Intermediate
Intermediate
Intermediate
Intermediate
Intermediate
See more

Language

English
Fluent
German
Intermediate

Work Experience

Research Associate - AI and Data Science at Carl von Ossietzky Universität Oldenburg
October 1, 2024 - Present
Working on the TaDeA (Tax Defence Analytics) project, building scalable machine learning pipelines on over 3 million records using Python and various databases including OracleDB and PostgreSQL. Developed graph-based machine learning models with Neo4j for anomaly detection and entity analysis. Successfully fine-tuned and deployed large language models such as BERT, GPT, Mistral, and Qwen, achieving a 15% accuracy improvement. Engineered retrieval-augmented generation pipelines using LangChain and FastAPI, containerized with Docker. Automated dataset collection of over 10,000 entries for training and applied MLOps practices including MLflow, CI/CD, and monitoring for reproducible deployments. Collaborated cross-functionally with data scientists, researchers, and product owners to influence key project decisions.
Master’s Thesis Researcher at RPTU (TU Kaiserslautern)
January 31, 2024 - August 26, 2025
Developed an encoder-decoder transformer architecture for automatic syntax generation, translating natural language to code using transfer learning. Built ETL pipelines for dataset creation and fine-tuned BERT and CodeBERT models using Hugging Face Transformers and PyTorch. Monitored training and evaluation with MLflow and deployed the model via REST API with real-time inference UI, achieving a CodeBLEU score of 97%, surpassing baseline GPT benchmarks.
Software Engineer at Rohde & Schwarz GmbH & Co. KG
December 31, 2023 - August 26, 2025
Contributed to the Creativity Platform Web Application project by documenting and evaluating technical requirements. Implemented frontend interfaces using ReactJS and developed end-to-end architecture with Docker, Kubernetes, CI/CD pipelines, and Microsoft Azure in an Agile environment.
Program, Project, and Process Management Associate at Accenture GmbH
October 31, 2022 - August 26, 2025
Conducted manual and regression testing using SOAP UI and MySQL, improving testing efficiency by 10%, particularly in regression phases. Documented and implemented test cases utilizing JIRA, XML, and JavaScript. Managed version control for test cases and cycles with GitLab.
Web Development Intern at LogWin Technology
October 31, 2020 - August 26, 2025
Developed responsive web interfaces using HTML5, CSS3, JavaScript, Bootstrap, and ReactJS.

Education

Master of Science at Rhineland-Palatinate Technical University of Kaiserslautern-Landau
April 1, 2021 - May 1, 2024
Bachelor of Engineering at Gujarat Technological University
August 1, 2016 - August 1, 2020

Qualifications

AWS Certified Machine Learning Specialty 2025 - Hands On – Udemy
January 1, 2025 - August 26, 2025
Generative AI with Large Language Models - DeepLearning.AI & AWS
August 1, 2023 - August 26, 2025
Natural Language Processing with Attention Models - DeepLearning.AI
May 1, 2023 - August 26, 2025
Sequence Models - DeepLearning.AI
April 1, 2023 - August 26, 2025
Machine Learning - Stanford University
December 1, 2021 - August 26, 2025

Industry Experience

Education, Software & Internet, Professional Services, Manufacturing, Government

Experience Level

Expert
Expert
Expert
Expert
Intermediate
Intermediate
Intermediate
Intermediate
Intermediate
Intermediate
Intermediate
Intermediate
See more