Research Assistant, 3 positions / Tutkimusapulainen, 3 paikkaa (Data Science Laboratory)

Research Assistant, 3 positions / Tutkimusapulainen, 3 paikkaa (Data Science Laboratory)

Tampere University and Tampere University of Applied Sciences create a unique environment for multidisciplinary, inspirational and high-impact research and education. Our universities community has its competitive edges in technology, health and society. www.tuni.fi/en

We are seeking master's degree students to work as Research Assistants on multiple different projects in the Data Science Laboratory at Tampere University. The positions are in the Computing Sciences unit of the Faculty of Information Technology and Communication Sciences.

JOB DESCRIPTIONS

At Tampere University, the title of Research Assistant is generally assigned to students who are both studying and working at Tampere University and perform research duties alongside their studies. Research Assistants are usually hired on a part-time contract, but they can also occupy a full-time position, for example, during the summer or while engaged in a thesis project.

We are now seeking Research Assistants to work on three specific topics in the Data Science Laboratory / Data Science Research Centre. Please, indicate in your application your top two preferred topics from the following:

1. Inference Optimization Strategies for High-Frequency Financial Time-Series Prediction

Supervisors: Juho Kanniainen and Alexandros Iosifidis

This project reduces inference speed of state-of-the-art machine learning models for high-frequency trading. The work studies speed-accuracy trade-offs focused on structured pruning suited to CPU/GPU dense computation. Candidate toolchains include Hugging Face Optimum (ONNX export, graph optimizations, quantization), Torch-Pruning (parameter removal and network reconstruction), NVIDIA Model Optimizer (8/4-bit precision and TensorRT integration), and PLiNIO (gradient-based compression). A three-step plan covers baselines, pruning pipeline prototyping, and real-world evaluation with a confe