The posting
About Ambee
Ambee is building climate intelligence. We combine proprietary on-ground sensor data, satellite and earth observation data, and advanced data science to produce hyperlocal insights on air quality, pollen, weather, wildfire and more. Businesses and organisations use this intelligence through our APIs and platforms to understand environmental risk and make better decisions. Behind it are data systems that process massive volumes of data every day, so scale, accuracy and reliability are core to everything we build.
About the role
We're hiring a Data Scientist to build the models and analytics behind Ambee's climate intelligence products. You'll turn sensor readings, satellite imagery and weather data into accurate, hyperlocal predictions for air quality, pollen, weather and wildfire risk. You'll own problems end to end, from exploring raw data and engineering features to training, validating and shipping models that power the APIs our customers rely on every day. The work involves large spatio-temporal datasets, noisy real-world signals and models that must hold up in production. You'll work closely with engineering, product and domain experts, and help shape how data science is done at Ambee.
What you'll do
- Work with stakeholders to frame business and scientific questions as well-defined data science problems with clear success metrics.
- Build, train and validate statistical and machine learning models for forecasting, nowcasting, spatial interpolation and anomaly detection across environmental data.
- Combine and clean data from ground sensors, satellites, weather models and third-party sources, handling gaps, drift, outliers and calibration issues.
- Engineer features from large spatio-temporal datasets, including time series, geospatial rasters and point data.
- Design rigorous evaluation: backtesting, cross-validation, error analysis and comparison against ground truth.
- Work with engineering to deploy models to production, and monitor their accuracy, drift and cost over time.
- Run exploratory analyses and experiments, and turn findings into clear recommendations for product and business teams.
- Build reusable pipelines, notebooks and tooling that make the team's work reproducible and faster.
- Stay current with research in machine learning, remote sensing and climate science, and bring the best ideas into our products.
- Document methods and results so engineers and non-technical stakeholders can understand and trust them.
What you'll need
- Bachelor's or Master's degree in Engineering in Computer Science, or a degree in a relevant field such as Statistics, Mathematics, Physics, Data Science, Atmospheric or Environmental Science.
- 1+ years of professional experience in data science or applied machine learning.
- Strong foundations in statistics, probability, linear algebra and machine learning.
- Strong Python skills and hands-on experience with pandas, NumPy, scikit-learn, and at least one deep learning framework such as PyTorch or TensorFlow.
- Experience with time series forecasting and modelling, using classical methods (ARIMA, state space models) and ML approaches (gradient boosting, neural networks).
- Good SQL skills and experience working with relational or NoSQL databases.
- Experience taking models from prototype to production, working with engineers on APIs, monitoring and retraining.
- Experience working with large, messy, real-world datasets and turning them into reliable insights, including optimising code, queries and memory use for speed and scale (vectorisation, chunking, partitioning, efficient formats like Parquet).
- Ability to design sound experiments and evaluate models honestly, including uncertainty and failure cases.
- Familiarity with Git, code reviews and writing clean, tested, reproducible code.
- Clear communication of complex results to technical and non-technical audiences.
Good to have
- Experience with big data technologies such as Apache Spark, Databricks, Hadoop, Dask or Kafka, and processing datasets that don't fit on a single machine.
- Experience building data pipelines (ETL/ELT, batch or streaming) with tools like Airflow or similar orchestrators.
- Experience with geospatial and remote sensing data (GeoPandas, xarray, rasterio, GDAL, NetCDF, GRIB) and satellite products.
- Background in atmospheric science, meteorology, air quality, hydrology or other environmental domains.
- Experience with cloud platforms, especially AWS (S3, SageMaker, EMR, Athena, Lambda).
- MLOps exposure including MLflow, Docker, model registries, CI/CD for models and observability.
- Experience with LLMs and generative AI frameworks (for example LangChain, LlamaIndex, OpenAI or Anthropic APIs, vector databases).
- Publications, Kaggle results or open-source contributions in relevant areas.
Who you are
- You work independently. You take open-ended problems, break them down and drive them to a shipped model with little supervision.
- You're rigorous. You question your data, test your assumptions and care about whether a model is actually right, not just whether the metric looks good.
- You think about production from day one: accuracy, latency, cost and how a model behaves when the data changes.
- You communicate clearly with engineers, scientists and business stakeholders alike.
- You're driven by the impact of your work, love digging into hard problems, and are always keen to learn something new.
Why join Ambee
- Work on real-world problems with large environmental datasets from sensors, satellites and weather models, used by customers worldwide.
- Build models that help people and businesses respond to climate risks.
- High ownership, a small and fast-moving team, and room to grow into technical leadership.
- Exposure to data engineering, AI/ML and cloud infrastructure alongside core product work.
Not quite this role?
We're always looking for people who bring new perspectives, curiosity and fearlessness to our work. Write to us at [email protected] with the role you have in mind and we'll get in touch.
