Thesis Work, 30 Credits - Knowledge-Graph Based Maintenance Agent in Pharmaceutical Manufacturing
Notify me about similar jobsAbout this opportunity
Are you interested in AI and pharmaceutical manufacturing? In this thesis project, you will develop and evaluate a knowledge-graph-based maintenance agent using real-world maintenance data.
About AstraZeneca:
AstraZeneca is a global, science-led, patient-centred biopharmaceutical company focusing on discovering, developing, and commercialising prescription medicines for some of the world’s most serious diseases. But we’re more than a global leading pharmaceutical company. At AstraZeneca, we're dedicated to being a Great Place to Work and empowering employees to push the boundaries of science and fuel their entrepreneurial spirit.
About the Opportunity:
As a Thesis Worker at AstraZeneca, you’ll find an environment that’s full of unique opportunities and exciting challenges. Here, you’ll have the opportunity to pursue your areas of interest whilst equally developing a broad skillset and knowledge base to get the best out of your experience. You’ll be working on meaningful projects to make an impact and deliver real value for our patients and our business.
Thesis work description:
Sustaining high uptime in pharmaceutical production depends on how quickly the right knowledge reaches the person standing at the machine. This knowledge comes from two complementary sources: the data generated by production and the accumulated understanding of how equipment behaves and how faults have been resolved in the past. A reliable agentic system needs access to both.
In this thesis project, you will focus on the second source: transforming historical maintenance knowledge into a knowledge-graph-based maintenance knowledge base that can be queried by an agent powered by a large language model (LLM). The data consists of historical maintenance records written as free text by technicians who carried out the work
The project builds on an existing information-extraction pipeline and agent prototype. You will investigate how maintenance records can best be transformed into a useful and scalable knowledge graph, and evaluate how effectively the LLM-driven agent can retrieve and use this knowledge when answering realistic maintenance questions.
Key objectives:
Verify the current knowledge graph construction pipeline and improve its underlying ontology against real maintenance records.
Investigate strategies for building an inductive knowledge graph, one that takes in new work orders, equipment, and fault types without being rebuilt.
Gather realistic maintenance questions from engineers and technicians and build them into an expert-reviewed evaluation set.
You will work with real manufacturing data in a regulated environment and collaborate closely with operators, maintenance personnel and technical experts. The project combines practical work in agentic AI and data engineering with knowledge representation and evaluation in an industrial setting.
Outcome:
You will gain hands-on experience with agentic AI built on real manufacturing data in a regulated environment and will deliver a working agent together with an evidence-based assessment of its reliability.
Placement:
This is an on-site position at AstraZeneca Södertälje.
Please note, AstraZeneca does not support with accommodations for this role.
Structure:
Duration: Spring 2027
Credits: 30
You will work together with another master’s thesis student who has already been identified for the project.
Essential Requirements:
Enrolled in a master’s programme in industrial engineering, computer science, artificial intelligence, production systems or a related field.
Strong interest in LLM-Ops, agent development and applied artificial intelligence.
Strong communication and collaboration skills, including the ability to interact with operators, engineers and subject-matter experts.
Experience with knowledge graphs or information retrieval is an advantage but not a requirement.
So, what’s next?
Apply today and take the chance to be part of making a difference, making connections, and gaining the tools and experience to open doors and fulfil your potential. We can´t wait to hear from you!
We welcome your application as soon as possible, but ahead of the scheduled closing date October 13, 2026. In the event that we identify suitable candidates ahead of the scheduled closing date, we reserve the right to withdraw the vacancy earlier than published.
Date Posted
22-sep.-2026
Closing Date
13-okt.-2026
Our mission is to build an inclusive and equitable environment. We want people to feel they belong at AstraZeneca and Alexion, starting with our recruitment process. We welcome and consider applications from all qualified candidates, regardless of characteristics. We offer reasonable adjustments/accommodations to help all candidates to perform at their best. If you have a need for any adjustments/accommodations, please complete the section in the application form.
Job details
How this role compares
Computed from every other active Manufacturing & CMC role in our database, not just this employer's listings.
We currently track 3720 comparable Manufacturing & CMC roles across 172 biopharma companies.
Salary context
906 of 3720 peers report a salary range (USD, annualized)
Peers share this role's job function. This posting doesn't list a seniority level, so peers aren't narrowed by seniority either -- the range below may span more levels than usual.
Where these roles are based
Top locations among the 3720 comparable roles
+ 58 more countries
Seniority mix
1647 of 3720 peers have a known seniority level
Therapeutic area mix
24 of 3720 peers have a known therapeutic area; the rest are genuinely unlabeled, not hidden
Similar opportunities
The closest matches from our peer group, ranked by how similar they are, not how well you'd qualify for them -- treat this as market context, not a guaranteed shortlist; a weak match is labeled as one below.
Notify me about similar jobs
Get an email when we spot other openings like this one – same job function, comparable seniority, roles you'd actually want to see.
Or follow a whole field: get every new Manufacturing & CMC job in Sweden, and choose locations and companies.
How we calculate "similar"
No black box, no LLM guesswork: a deterministic score built from four normalized attributes. Here's this role's own peer group at different match levels, so you can see the mechanism, not just the result.
Every comparison starts from the same 100-point budget: 25 for working in the same function, 40 for the same therapeutic area, 20 for the same or adjacent seniority, 15 for the same country. A dimension we can't confirm on both sides contributes nothing, never a guess, never a free pass.
0 points, never a partial guess. A role we know almost nothing about beyond its function bottoms out at 25%; it never inflates to 100% just because there's little to compare against. Seniority uses a defined ladder (Associate → Manager → Associate Director → Senior → Principal → Director → Senior Director → Executive/VP) so "Director" and "Senior Director" count as adjacent, but "Director" and "Executive/VP" do not.