GPT-Lab · Tampere University
Master Thesis Topic Bank
Welcome to the Master Thesis Topic Bank at GPT-Lab! This resource provides a curated list of thesis topics for master's students at Tampere University who wish to conduct their thesis in the fields of artificial intelligence and software engineering. Here, you’ll find a variety of predefined topics that reflect the cutting-edge research happening at GPT-Lab. Each topic is designed to align with our core focus areas, providing you with opportunities to contribute to innovative projects.
You are encouraged to select a topic from the list or propose your own in the registration form, provided it falls within the scope of AI and software engineering. Whether you’re passionate about natural language processing, machine learning, or software development, you’ll find exciting opportunities that match your academic and research interests.
For those proposing their own topics, please ensure they meet the criteria and are relevant to our ongoing research themes.
How it works
Browse Topics
Review the list of available thesis topics, each with a brief description.
Select or Propose
Choose a topic or propose your own idea during the application process.
Get Started
Once your application is accepted, you will be paired with an advisor to begin your research journey.
Open the registration form
Scroll down to browse the available topics.
Disclaimer: For the best viewing experience, please access this table on a desktop device.
Last updated: 07.10.2026
| Available | A Lightweight Command Line GPU Time Allocation System for a Multi-Organization Shared Server | Jussi Rasku | This project aims to design and implement a lightweight, single server, command-line-based GPU allocation system used by multiple organizations. The system will enable users to reserve GPU time efficiently, show the reservations, and prevent use of unallocated resources, ensuring fair usage across various projects. The student will also compare the solution against existing systems in terms of performance and ease of use. | Strong programming skills (preferably C), knowledge of Linux systems and command line interfaces, and experience with distributed computing environments. |
| Available | Adaptive Language Models for Game Rule Generation and Quality Evaluation | Jussi Rasku | This thesis will explore how language models can be adapted to generate and evaluate novel game rules in the context of both card and abstract board games. Inspired by Browne and Maire’s work on evolutionary game design, the research aims to create a simulation environment where language models generate game rules and assess the “quality” of these games based on parameters like drama, uncertainty, and playability. The models will be benchmarked against a suite of metrics to optimize for engaging gameplay, informed by principles from recent studies in game design and AGI experimentation (e.g., ARC-AGI, AlphaCode). | Proficiency in LLM use and APIs, experience with Python, and an interest towards evolutionary algorithms. Familiarity with game design concepts. |
| Available | AI-BASS Real Options Subsystem for Startup Decision Modeling | Jussi Rasku and Mikko Auranen | Objective: To design and implement a real options subsystem integrated into the AI-BASS platform to support startup-level investment and resource allocation decisions. The system will detect relevant decision situations during AI-BASS chat sessions, model them as real options, and provide an interactive visualization of active options and their dependencies. Methodology: This is a constructive research to develop a microservice-based subsystem triggered within the AI-BASS chat when an investment or allocation discussion arises. The subsystem will: (1) query contextual data from connected sources (company documents, emails, and prior chats stored within AI-BASS instance); (2) structure decision information and store it into a MongoDB schema designed for real options representation; (3) generate visual graph-based representations of active options and dependencies. Experimental validation will be conducted using synthetic and pilot company data to assess the subsystem’s ability to identify and model real options. As an optional extension, a lightweight code interpreter tool component may be explored to enable the AI agent to perform preliminary real options valuation and its uncertainty. Rationale:This research bridges AI-assisted decision support and real options analysis for startups, addressing operational-level uncertainty modeling rather than venture capital valuation. It provides a concrete technical contribution to the AI-BASS ecosystem and forms a strong foundation for the student to work on applied system development, decision-theoretic modeling, and experimental evaluation. | The student should be fluent in Python, LLM APIs, and comfortable with microservice architecture and MongoDB. They must understand and willing to learn of basic decision analysis and real options theory, and be able to translate these into computational models. Experience with data integration, network models, visualization frameworks, and applied AI systems is essential. The work requires technical autonomy, analytical rigor, and the ability to validate and document a novel subsystem through prototype development. |
| Available | AI-Supported Strategy in Software Companies: A Qualitative Mapping of Practices and Implications | Jussi Rasku and Mikko Auranen | Objective: To examine how AI is currently reportedly used in developing and implementing strategy in software companies, and how this relates to decision-making quality and software/process outcomes. Methodology: 1. Collect academic and non-academic sources discussing AI use in strategic processes. 2. Conduct thematic content analysis to identify how AI tools are applied in strategic planning, communication, monitoring, and adaptation. 3. Then synthesize a conceptual framework linking AI-supported strategic activity with decision-making style and outcome quality. Here, use illustrative case material from publicly available examples. Methodologically this could be a systematic literature review or grounded theory based research. Rationale: Enables grounded insight into emerging practice without requiring access to internal data. Suitable for sparse formal literature. | The student should understand strategic management and AI in business, be able to conduct systematic or grounded literature analysis using qualitative or text analysis tools such as NVivo, Atlas.ti, or Python, and be capable of synthesizing findings into a conceptual framework. Strong academic writing skills and independent analytical thinking are essential. |
| Available | Automated Software Quality Assurance using LLM | Shahbaz Siddeeq | To improve this, real-time integration with tools like Slack, GitHub, and CI/CD pipelines allows the system to react quickly when requirements or code change. Agents can monitor communication channels, track code updates, and assess how these changes impact other parts of the project. This ensures that the system is always up-to-date and can trace how a change in one area affects other areas, such as tests or design documents. | Strong programming skills, software testing proficiency, LLMs, and AI/ML knowledge. |
| Available | Can I Trust This Model? A Study on How Documentation Shapes Trust in Open-Source AI | José Siqueira de Cerqueira | This thesis explores how open-source AI model documentation — especially model cards — influences developers’ trust in the models themselves. As developers rely on documentation to assess risks and reliability, the study investigates which elements (e.g., training data, use cases, benchmarks, limitations, licensing) most impact perceived trust. Through semi-structured interviews, it will assess how developers interpret documentation. The study also examines whether transparency and explainability increase model trust, and whether developers distinguish between trusting the documentation versus the model. Findings will inform better documentation practices to support trustworthy AI development. | Familiarity with how ML/LLM models work (training, evaluation, datasets, etc.) Willingness to learn or experience with interviews, thematic analysis, or survey design |
| Available | Comparative study of Finnish capable Large Language Models over several software engineering tasks | Jussi Rasku | The thesis will focus on evaluating language models fluent in Finnish such as Poro and Viking across various tasks related to sofware engineering. They are compared against the state-of-the-art commercial offering such as ChatGPT models. The student will benchmark the models in both CPU and GPU environments and assess their performance and accuracy. | Familiarity with large language models, natural language processing and machine learning concepts. Experience with benchmarking tools and Hugging Face models is recommended. |
| Ethical issues in AI agents and multi-agent collaboration: a systematic literature review | Kai-Kristian Kemell | As AI continues to progress, hypothetical future risks become the reality of today. AI agents, both lone agents and AI agents working collaboratively, present various promises across a wide variety of industries, but also various risks. Discussion on these systems has accelerated with recent advances in Generative AI (GenAI) and Large Language Models (LLMs). | TBA | |
| Available | From RPA Code to BPMN Process Diagrams | Jussi Rasku | This thesis investigates methods for automatically transforming robotic process automation (RPA) code into BPMN process diagrams to improve process transparency, maintainability, and integration with business process management tools. The work involves analysing RPA scripts, defining mappings to BPMN constructs, and prototyping a converter. The topic comes from Seinäjoki-based IT-consultation company River IT and details of collaboration can be agreed upon. | Prerequisites include solid programming skills, understanding of RPA platforms and BPMN modelling, and ability to read and formalise code-to-model transformations. |
| Available | Gameful LLM-Based Multi-Agent Systems | José Siqueira de Cerqueira | This thesis aims to assess what gameful elements can effectivelly influence a Multi-Agent system’s performance in coding tasks. By introducing gameful elements in a Multi-Agent system, we hope to encourage collaboration (or competittion), leading to more refined output. | Familiarity with how ML/LLM models work (training, evaluation, datasets, etc.) Willingness to learn or experience with interviews, thematic analysis, or survey design |
| Available | Impact of generated PlantUML diagrams to large language model generated code accuracy | Jussi Rasku | The performance of a LLM generating or fixing code is dependent on the correct context in which the edits are made. Hence, the research question is: Can the performance of the coder AI be improved by enriching the prompt with generated UML diagrams? The diagrams are to be generated from relevant project source files and given to the LLM using a suitable DSL such as PlantUML. The feasibility of the approach is tested on SWE- bench, some subset of that, or some other coding benchmark. | The student has experience in using coding LLMs such as Github Copilot, Qwen 2.5, Mistral or DeepSeek Coder. Also, one should be fluent in creating and reading UML, know how to write PlantUML, and be interested in empirical testing of software tools. |
| Available | Smart and Sustainable Energy Consumption Optimization with IoT and LLMs | Mika Saari | Are you interested in developing AI-powered solutions for energy efficiency? This Master’s thesis project focuses on creating an IoT-based energy management system where Large Language Models (LLMs) analyze and optimize energy consumption. The project could involve integrating energy meters and IoT sensors, such as temperature, humidity, and electricity usage monitors. It could also include developing AI-driven analytics to provide smart recommendations for optimizing energy consumption. Additionally, a user-friendly interface could be created for energy management and forecasting. By reducing energy consumption and improving decision-making, this project contributes to sustainability and the circular economy. If you are passionate about AI, IoT, and green technology, this is your chance to make an impact. | Python, LLM. IoT -devices Possibility to use/implement APIs: https://github.com/Green-Software-Foundation |
| Available | Sustainable Software Engineering with AI | Mika Saari | Examples of possible topics: “Reducing the Carbon Footprint of AI Model Training: A Study on Pruning, Quantization, and Transfer Learning”, “Life Cycle Assessment of AI Models: From Training to Deployment”, “AI-Specific Hardware for Green Computing: A Comparative Study of GPUs, TPUs, and AI Chips” | https://github.com/Green-Software-Foundation |
| Available | From Retrieval to Context Engineering: Task-Aware Context Construction for Agentic RAG | Toufique Hasan | This thesis studies how Agentic RAG systems can go beyond simple top-k document retrieval. Instead of only selecting the most similar chunks, the system builds context based on the user’s task, intent, retrieval needs, and expected answer. The study focuses on task understanding, retrieval planning, context selection, filtering, and answer verification before producing the final response. This thesis studies how Agentic RAG systems can go beyond simple top-k document retrieval. Instead of only selecting the most similar chunks, the system builds context based on the user’s task, intent, retrieval needs, and expected answer. The study focuses on task understanding, retrieval planning, context selection, filtering, and answer verification before producing the final response. | Python, basic RAG knowledge, LLM agents, vector databases, prompt engineering, basic evaluation methods |
| Available | Reverse Engineering Requirements Specifications from Existing Software Products | Risha Parveen, Zheying Zhang | This thesis investigates how requirements specifications can be reconstructed from an existing software product when the original requirements are missing, incomplete, or outdated. The work will explore what information can be extracted from sources such as the software itself, user interfaces, documentation, source code, tests, and other available project artifacts, and how this information can be used to recover functional and non-functional requirements. The thesis may include the development and evaluation of an AI-assisted approach for generating a structured requirements specification from an existing product. | Basic knowledge of requirements engineering and software engineering. Programming skills and familiarity with LLMs or AI-assisted software engineering are beneficial. |
| Available | Improving Trustworthiness and Traceability in AI-Generated Product Vision Documents | Risha Parveen, Zheying Zhang | This thesis investigates how AI-assisted generation of software product vision documents can be improved to produce more trustworthy and traceable results. The work will explore approaches for ensuring that generated content is grounded in the available project information, that important statements can be traced back to their supporting sources, and that unsupported or incorrect information is reduced. The thesis will involve improving an existing approach and evaluating the quality, trustworthiness, and traceability of the generated vision documents. | Basic knowledge of software engineering and requirements engineering. Programming skills and familiarity with LLMs, prompt engineering, or retrieval-based AI systems are beneficial. |
| Available | Barriers and enablers of AI governance in organisations | Salla Westerstrand | The goal of the thesis is to seek understanding of barriers and enablers that organisations face when trying to govern AI in a meaningful way. | Basic knowledge of responsible AI and AI governance. The most important is having interest towards understanding organisational aspects of AI adoption and how it can be done in an ethically and societally sustainable way. |
| Available | LLM-Coordinated Multi-Agent Reinforcement Learning with Surrogate Models or Digital Twin for Industrial HVAC Optimization | Rajratan Wankhade | The study will use a digital twin or surrogate-model-based HVAC simulation environment to represent a multi-zone industrial facility.The selected simulation approach will provide an efficient environment for multi-agent reinforcement learning and HVAC optimization. An LLM-based supervisory agent will be investigated for high-level coordination of the RL agents according to production conditions, energy constraints, and environmental requirements. | Basic knowledge of ML & RL algorithms, Python programming, and prompt engineering is expected. Familiarity with HVAC systems is beneficial but not required, as the work will primarily focus on AI agents, ML/RL-based optimization, and agent orchestration frameworks such as LangGraph. The use of a digital twin or a surrogate model as the simulation environment will be determined based on the final research scope and available resources. |
| Available | Small Language Models for agent memory management systems. | Mahade Hasan | Existing studies (e.g., MemGPT, Mem0, A-MEM) demonstrate different memory mechanisms, but they typically treat a LLM as the agent’s “brain.” The capabilities of Small Language Models (SLMs) in this role remain empirically understudied, even though recent work claims SLMs are a natural fit for agentic systems. It is therefore important to understand how well SLMs perform in memory management compared to LLM baselines, and whether using SLMs introduces any trade-offs. Goal: 1. Empirical Data Collection: – Replace LLM with SLM in the existing Memory management systems (e.g.; MemGPT, MemZero, Zep, A.Mem) and evaluate them with established benchmarks (e.g.; LoCoMo, LongMemEval). 2. Data Analysis: – Analyze the data with statistical methods and find patterns from the data. 3. Report and Interpret: – Report the results with the trends. – Interpret the result to find and understand the key insights. | Solid understanding of Python programming.Familiarity with agentic systems and memory mechanism is a plus, if not the student will learn on the go.Also, some working experience with agentic frameworks like langchain, crewai, langgraph is appriciated. |
| Available | Small Lanaguage Model based Multi-agent system for use prompt analysis. | Mahade Hasan | This study will follow a design science research approach where the student will get the chance to work with Multi-agent Systems (MAS) using Small Language Models. The goal is to understand how we can create a dynamic dashboard from the real customer data to understand their behaviour around the system. | 1. Solid understanding of Python programming 2. Familiarity with agentic systems and some working experience with agentic frameworks like langchain, crewai, langgraph. 3. Basic knowledge of LLM to SLM conversion such as, quantization, knowledge distillation, Finetuning is a plus. |
| Available | An Ethical AI Study in the Context of the EU AI Act: Human Oversight and Accountability for Agentic AI and Its Generated Artifacts | Chalisa Veesommai Sillberg | AI agents and agentic AI systems can autonomously plan and perform tasks and create or modify software artifacts, such as requirements, source code, test cases and documentation. This increasing autonomy raises questions about human oversight, ethical responsibility and accountability. This thesis investigates these questions from an Ethical AI perspective, using the EU AI Act (particularly its human-oversight provisions) as the regulatory context. Goal: The goal is to identify the main ethical and accountability challenges associated with agentic AI and examine how professionals understand appropriate human monitoring, approval, intervention and responsibility. Method: The research will be conducted in two stages: 1. Systematic scoping review: Relevant academic studies on Ethical AI, agentic AI, human oversight and accountability will be systematically collected and analysed. 2. Scenario-based interviews: AI or software professionals will be interviewed in which an AI agent creates or modifies software artifacts with different levels of autonomy. The interview data will be analysed using thematic analysis to identify common views, concerns and oversight expectations. | 1. a background or basic understanding in software engineering, computer science, information technology, AI or a related discipline 2. a basic understanding of generative AI, AI agents or autonomous AI systems; 3. an interest in Ethical AI, responsible AI and the EU AI Act |