Applied Machine Learning Lead

About Repli5: Repli5 is a small Swedish startup based in Gothenburg that specializes in bringing Generative AI to computer vision applications. Our technology, called RealSim, leverages gen AI to bridge the gap between synthetic data and reality. This is particularly useful for creating accurate, realistic training data for computer vision models, essential in developing solutions for industrial robotics, autonomous drones, and select automotive applications. Our technology is also capable of enhancing real-world images by introducing extensive weather and lighting variations, which significantly increases the robustness of downstream computer vision models. We are now packaging this tech into a platform service built for computer vision engineers. The Opportunity: Shaping the Future of AI Perception We're looking for an experienced Machine Learning Engineer with a deep background in computer vision and generative AI. You’ll be part of a team developing workflows that generate realistic, diverse, and highly controllable synthetic image data using state-of-the-art tools like Stable Diffusion, LoRA, and ControlNet. These workflows will be adapted and applied to our new platform and your work will thus directly impact Repli5’s business and the performance and robustness of real-world AI systems—from self-driving cars to industrial automation. What You’ll Do (Key Responsibilities) Develop and fine-tune generative AI pipelines (e.g., Stable Diffusion, ControlNet, LoRA) for synthetic image generation. Design and implement machine learning workflows to generate structured, photorealistic datasets. Collaborate with 3D artists and simulation engineers to align ML models with physical environments and sensor setups. Work closely with customers in automotive, robotics, and aerospace to deliver tailored datasets. Optimize generation systems for scalability, quality, and speed using tools like ComfyUI or custom pipelines. Who You Are (Qualifications) The Essentials: MSc or PhD in Computer Vision, Machine Learning, or related field with 3+ years of hands-on experience with deep learning frameworks (e.g., PyTorch, TensorFlow). Extensive experience with state-of-the-art image generation models (GAN, Diffusion, Transformers) Exploring Mindset: Experience exploring workflows with generative AI models (especially diffusion-based) and a strong passion for reading and applying new research papers. Toolset Curiosity (Big Plus!): Familiarity with tools like LoRA, ControlNet, ComfyUI, or custom workflows around Stable Diffusion. Domain Knowledge (Big Plus!): Strong understanding of synthetic data generation and its role in training perception systems. Practitioner: A track record of deploying ML systems in domains like industrial automation, defense, or automotive. The Startup Gene: You excel in fast-paced, ambiguous environments where you can take initiative and work independently. A Team Player: A problem-solving intellect with a team player spirit and a positive can-do attitude. What We Offer A competitive compensation package. Flexibility and work-life balance. Challenging tasks that are at the forefront of technology. The autonomy to make major technical decisions and influence the product direction. An international, caring culture where it is natural to both ask for and offer help. Flexible working hours and the possibility to work from home some days a week. Fun at work. Recruitment Process Application Review First Interview (1-1 with COO) - More about you. This is our chance to learn who you are and what you bring to the table. Technical interview (with 2-3 from the Repli5 team). You're given a technical problem similar to a customer request and you are expected to present how you would solve this (but not present the actual solution). This is followed by general technical questions Final stage is a sync with the CEO.

Turn ship physics into software people actually use At Cetasol we build the monitoring and analytics platform that commercial fleets use to cut fuel consumption, understand how their vessels really perform, and stay on the right side of CII and EU MRV reporting. A growing part of that platform is predictive maintenance: catching when an engine, generator, thruster, or the vessel as a whole starts behaving differently than it should, before it becomes a failure. Getting there takes two things that rarely live in the same person: an understanding of the physics behind how these assets actually behave, and the engineering skill to ship that understanding as a working product. We're looking for that person. What you'll do Build physical and data-driven models of vessel assets — engines, generators, thrusters, and the ship as a whole — that capture what "normal" operation looks like Detect when an asset deviates from its expected behaviour, and trace that deviation back to the sensors and signals driving it Work with noisy, real-world sensor data from vessels: clean it, validate it, and decide what it can and cannot support Take your own models from notebook to production on AWS, and build the React interfaces where the results actually land in front of customers Own features end to end, with our UX designer, full stack developers and data scientist alongside you Help us decide where a first-principles model beats a statistical one, and where it doesn't What we're looking for The balance here is genuinely 50/50. You should be comfortable deriving a model and deploying it. On the modelling side: Strong applied mathematics, physics or engineering foundation — fluid dynamics, thermodynamics, control theory, numerical methods, or similar Experience with anomaly detection or condition monitoring on physical systems — knowing how to separate a genuine fault signature from normal operating variation Solid Python and the usual scientific stack; experience with time series and sensor data The judgement to know when a model is good enough and when it's fooling you On the engineering side: Practical experience building and running services on AWS Comfortable in React and modern frontend work — you don't need to be a specialist, but you should be able to build the view your model deserves Familiarity with the ordinary craft of software: version control, testing, code review, CI And: You explain technical work clearly to people who aren't specialists You're happy to sit somewhere between disciplines rather than in the middle of one We're open on years of experience. If you can do the work, the title follows. Nice to have Predictive maintenance or condition-based monitoring · maritime, automotive, aerospace or energy domain experience · digital twins or physics-informed machine learning · edge computing or embedded data collection Why Cetasol Shipping accounts for roughly 3% of global CO₂ emissions, and a meaningful share of that is avoidable through better operational decisions. The work here is technically interesting and the impact is measurable in tonnes of fuel. You'd be joining a small engineering team where your work is visible and your scope is wide. We're based in Gothenburg and work hybrid — we value time in the same room, without counting days.