Senior Machine Learning Engineer

Role: Senior Machine Learning Engineer

Location: Berlin, Germany

Humankind has less than 6 years left to limit average global surface temperature rise to 1.5 degrees. This job offer constitutes an unique opportunity to prevent orders of magnitude of carbon emission, much more than you can achieve as an individual consumer or in most other forms of employment.

alcemy was founded to reduce the massive CO2 footprint of the cement and concrete industry. Concrete is humanity's second most used material by volume and is responsible for around 8% of global CO2 emissions.

We can change this! We leverage machine learning to provide smart production control software to the concrete supply chain to improve efficiency and allow for the production of low-carbon cement and concrete at large scale. Our software is used 24/7 in cement and concrete plants to optimize production and reduce carbon emissions today. We are currently enabling our customers to save more than 100 000 tons of CO2 emissions a year and we aim to save 100 million tons per year in 2030.

THE OPPORTUNITY

You will be at the heart of our cement product, working on our machine learning pipeline. Our machine learning system is running in production in 10+ cement plants and is taking control over the milling process of cement. That comes with a huge responsibility. Prediction accuracy must be high and the pipeline robust to data drifts as well as outliers.

From the beginning on, you will share ownership of the ML pipeline to work on all parts including data preprocessing, training and evaluating models, as well as the actual cement production control part.

As a Senior ML Engineer at alcemy, you will

• be responsible for the prediction accuracy of our models, with regards to its application in a production control loop
• run experiments and try to continuously improve our machine learning pipeline
• extend our ML monitoring to enable us to detect ML problems automatically
• share your experience and expertise with the team, by
  • mentoring more junior engineers
  • helping non-ML engineers to understand our ML product and its intricacies
  • serving as a sparring partner for other ML engineer
• deeply understand the domain of cement production to:
  • Make sure that our models are set up to make our customers successful
  • Make sure we develop processes with and for our customers to improve data quality
  • Develop new machine learning products for parts of the cement production that are not yet covered
• reduce the manual ML maintenance that is required to ensure the models run correctly, to enable us for the next phase of scale-up
• align with the solution engineering and customer success team about ML settings and customer specific changes

At alcemy, we use Amazon Web Services (AWS), which enables us to roll out our product at a fast pace. Our infrastructure is managed in terraform and everything runs on Kubernetes. The backend and machine learning tasks use Python, FastAPI, and PostgreSQL, as well as Python's numeric and data science ecosystem (numpy, pandas, scikit-learn, etc.). The frontend uses NextJS and Typescript. Furthermore, we emphasize proper software engineering practice, like CI/CD, merge reviews and testing to ensure a working production environment.

Requirements

YOUR PROFILE

• Experience in developing machine learning models that are used in production, including data preprocessing and feature engineering
• Comfortable with both, the data science and the ML engineering tech stacks (python, pandas, numpy, sklearn, SQL)
• Experience in software engineering, including working on a production code base, conducting merge reviews, and overseeing the deployment of one's own software
• Experience with developing and optimizing machine learning models to solve customer problems
• Clear grasp of experimental methods, experience performing experiments with real-world ML systems and real world data
• Comfortable creating visualizations to bring across complex data driven insights
• You have a high willingness, curiosity and growth mindset to learn and dive deep into the challenging cement and concrete domain to make our models excel in a real environment
• English conversational
• It is necessary that you are located in Germany. Alternatively, you should be willing to relocate to Germany and we will support you with the relocation

Even better:

• Experience in dealing with timeseries and/or tabular data
• Having built a machine learning system in a process industry context
• Experience with online learning use cases
• Experience with ML Ops tools and model deployment in general

WITHIN 1 MONTH YOU WILL

• Complete our onboarding and meet every team member.
• Learn a lot about cement, concrete and how our customers think. In addition, you will learn the intricacies of cement data and the production process.
• Get a deep dive in the product and understand all the nitty gritty details about producing cement to a good quality.
• Deploy your first changes to production.

WITHIN 3 MONTHS YOU WILL

• Have shipped first model accuracy improvements
• Extended our ML monitoring dashboards in Grafana
• Have met all your colleagues at our in-person team weeks
• Have a deep understanding how cement manufacturing and the resulting data are related to our machine learning setup

WITHIN 6 MONTHS YOU WILL

• Have setup a ML monitoring and alerting strategy such that we detect problems with our models easier and quicker
• Improved prediction accuracy across all cement plants
• Improved our current solution for how to deal with concept drifts in the cement data
• Laid out a roadmap for how to improve our ML pipeline over the course of the next year

Benefits

• We highly value transparency and direct feedback, and are always open to refining how we work together
• Our engineers take full ownership of their features, from implementation and testing through code reviews and deployment
• We have no on-call policy for our engineers
• We utilize modern communication tools such as Google Workspace and Slack
• We have been working diligently to establish a clear career progression framework
• Our team setup is a hybrid, meaning we have an office in Berlin, but we also provide the ability to work remotely. To ensure human connection, we have a quarterly on-site team week. During this time, we try to focus less on day-to-day work and more on evening activities, goal setting, and strategy definition

Our hiring process is fully remote and we’ll communicate with you over email and video chat.

• 1:1 screening call with one of our engineers
• Complete a take-home task
• A team fit interview (45 minutes) with 2 of your future team members
• A technical interview (90 minutes) with 2 of our engineers

We encourage applicants from all backgrounds to apply. We also acknowledge how stressful interviews can be. Let us know if there is anything we can do to improve the process so you can demonstrate your skill set.

Note: We currently do not offer visa sponsorships, so you have to have a valid work visa for the EU.

Tags