Looking for a job

Parham Nooranbakht

@parhamnooranbakht

Machine learning research engineer at Fraunhofer ISE fine-tuning foundation models on sensor data and improving prediction accuracy by 12%.

Germany

Message

At Fraunhofer Institute for Solar Energy Systems ISE, I fine-tune TimesFM, MOMENT, and TabPFN on multi-year sensor time-series data. I’ve achieved a 12% improvement in ML prediction accuracy and use Slurm for cluster-based compute jobs.

My master’s project at Albert-Ludwigs-Universität Freiburg developed SGQN, a value-guided adaptive discretisation method for continuous control. Graded 1.0, the project forms the basis of a paper submitted to TMLR and under review.

Before that, I designed Java and Spring Boot microservices for Khebreh Pardaz Pars (KPP), supporting 50,000+ daily API requests, and implemented REST APIs that reduced integration time by 40%. I also contributed to TabPFN research at the University of Freiburg Machine Learning Lab.

Experience

Work history, roles, and key accomplishments

Current

Machine Learning Research Engineer

Current

Fraunhofer Institute for Solar Energy Systems ISE

Mar 2025 - Present (1 year 7 months)

Fine-tunes foundation models on multi-year sensor time-series data, implements baseline models, and performs systematic hyperparameter optimization. Achieved a 12% improvement in ML prediction accuracy.

Python PyTorch TimesFM Moment TabPFN Hyperparameter Optimization Slurm Signal Processing Feature Engineering

Machine Learning Research Engineer

University of Freiburg, Machine Learning Lab

Jul 2024 - Dec 2024 (5 months)

Contributed to TabPFN research, focusing on improving the data preprocessing pipeline.

Python PyTorch TabPFN data preprocessing

Java Software Engineer

Khebreh Pardaz Pars (KPP)

Mar 2023 - Mar 2024 (1 year)

Designed a microservices architecture in Java and Spring Boot for 50,000+ daily API requests. Implemented REST APIs, established CI/CD pipelines, and optimized MS SQL Server queries.

Java Spring Boot Microservices REST APIs Docker Kubernetes CI CD MS SQL Server

Research Assistant - Deep Reinforcement Learning

Sharif University of Technology

Sep 2022 - Feb 2023 (5 months)

Implemented and evaluated deep RL algorithms (A2C, PPO, TRPO) in simulated robotic locomotion environments, achieving a 78% success rate. Accelerated training convergence by 30% through systematic environment modifications.

Python PyTorch Deep reinforcement learning A2C PPO TRPO Robotics

Research Assistant - Deep Learning

K. N. Toosi University of Technology

Jul 2022 - Nov 2022 (4 months)

Researched deep learning architectures and optimization methods; implemented neural networks in TensorFlow and PyTorch.

Python TensorFlow PyTorch Deep Learning Neural Networks

Education

Degrees, certifications, and relevant coursework

Albert-Ludwigs-Universität Freiburg

Master of Science, Computer Science

2024 -

Grade: 2.0

Activities and societies: Master's Project (18 ECTS) with grade 1.0: 'Discrete Q-Learning Approaches to Continuous Control', development of SGQN and basis of TMLR submission.

Pursuing a Master of Science in Computer Science with a specialization in Artificial Intelligence. Expected to complete in approximately 2 months, with a GPA of 2.0 and 84/120 ECTS completed.

K. N. Toosi University of Technology

Bachelor of Science, Computer Engineering

2018 - 2022

Grade: 19.07/20

Activities and societies: Dean's List 2020; top 3% of class (76 students).

Completed a Bachelor of Science in Computer Engineering with a GPA of 19.07/20, ranking in the top 3% of the class. Recognized on the Dean's List in 2020.