Senior Data Scientist

Nearmap's Data Science team designs, prototypes, and develops machine learning products using Nearmap imagery data. As a Senior Data Scientist (Model R&D), you will own the statistical and practical performance of models that ship – framing problems, selecting modelling approaches, training and validating models, identifying failure modes, and improving results through modelling choices, data strategy, and disciplined evaluation. Off-the-shelf models, with minor fine tuning often do not suit our needs, so expect to be making complex decisions about problem formation/specification and model architecture. You will work closely with Machine Learning Engineers who build the training and deployment infrastructure. Your core accountability is the quality, robustness, and suitability of the predictions those tools produce.

Important: This is a hands-on modelling role focused on novel approaches to computer vision and deep learning. If your strength is dashboards, BI, reporting, classical modelling, or deployment of off-the-shelf models, this role is not the right fit for you.

What You’ll Do

Own the modelling lifecycle end-to-end: problem formulation → dataset strategy → training → evaluation → error analysis → improvement

• Develop, or select and make novel adaptations to, model architectures for 2D/3D vision tasks using Nearmap’s multi-date, multi-angle, and 3D data
• Use PyTorch and important associated packages and libraries (timm, xformers) to develop bespoke models that underlie Nearmap products.
• Remain familiar with core conceptual elements of modern deep learning (attention, positional encoding, receptive field, pre-training), so that you can develop and adapt models to fit our needs.
• Keep up to date with important developments in the field, such as innovations in the fields of vision and 3D
• Design experiments and ablations to understand what drives performance and to de-risk model changes
• Improve model performance through sampling strategies, labelling guidance, quality checks, and targeted collection for hard/rare cases
• Work with ML Engineers to productionise models and ensure they are robust (monitoring, regression testing, reproducibility)
• Communicate clearly about trade-offs, failure modes, and performance, primarily to technical stakeholders

Required

• Background in a relevant discipline (Computer Science, Mathematics, Physics, or similar)
• Strong grasp of core ML fundamentals and the ability to explain them clearly (validation, regularisation, optimisation, over/underfitting, error analysis)
• Hands-on experience training and evaluating ML models end-to-end, including designing baselines and iterating based on evidence
• Strong coding ability in scientific Python in a Linux environment, using Git for source control
• Demonstrated scientific rigour: forming hypotheses, designing experiments, and interpreting results using appropriate metrics and statistical reasoning
• Experience delivering outcomes in real-world settings (shipping models, improving model quality over time, or delivering robust research prototypes)

Desirable

• Computer vision experience (segmentation, detection, representation learning)
• Experience working with geospatial data, aerial/satellite imagery, multi-view imagery, or multi-temporal imagery
• Experience with 3D vision / photogrammetry / structure-from-motion, or working with point clouds / 3D meshes
• Experience improving models via data strategy (label quality, sampling, active learning, hard-negative mining, dataset auditing)
• Experience training at scale (GPUs, distributed training) and working with large datasets that require efficient IO and compute
• Cloud computing experience (AWS or GCP)
• Software engineering skills for contributing production-quality code to shared codebases (testing, reproducibility, documentation)

Personal Attributes

• Curiosity and learning mindset: you keep your fundamentals fresh and stay up to date with modern deep learning
• Pragmatism: you are capable of novelty and creativity, but not for its own sake: you prioritise solutions that work and ship over novelty
• Strong judgment and “model taste”: you can choose sensible baselines, interpret failures, and iterate effectively
• Collaboration: you communicate well, share knowledge, and work effectively in teams

Some of our benefits

Nearmap takes a holistic approach to our employees’ emotional, physical and financial wellness. Some of our current benefits include:

• Quarterly wellbeing day off - Four additional days off annually for your 'YOU' Days
• Wellbeing and technology allowance
• Annual flu vaccinations
• Hybrid flexibility for this role 
• Nearmap subscription (of course!)
• Work from Anywhere policy for up to 4 weeks each year
• Stocked kitchen with access to all the snacks you need
• In-office catered lunch every Tuesday and Thursday at our Sydney CBD office
• Showers available for anyone cycling to work or lunchtime gym-goers!

Working at Nearmap
We move fast and work smart; often wearing multiple hats. We adapted to remote working with ease and are continually looking at ways to improve. We’re proud of our inclusive, supportive culture, and maintain a safe environment where everyone feels a sense of belonging and can be themselves.

At Nearmap, we embrace a flexible hybrid approach that empowers teams to determine what works best for them. Rather than mandating specific office days, we trust our teams to collaborate with their managers and decide when in-person time adds the most value.

This means you'll have the flexibility to balance remote work with office collaboration in a way that suits your role, your team, and your life. There are no company-wide mandatory office days, giving you the autonomy to work where you're most productive while staying connected with your colleagues.
If you can see yourself working at Nearmap and feel you have the right level of experience, we invite you to get in touch. 

Nearmap AI

• Read the product documentation for NearmapAI:  https://docs.nearmap.com/display/ND/NEARMAP+AI
• For a deep dive into Nearmap AI, listen to AI Systems Senior Director Mike Bewley on the Mapscaping podcast:  https://mapscaping.com/blogs/the-mapscaping-podcast/collecting-and-processing-aerial-imagery-at-scale

Note to agencies: Thanks, but we got this! Nearmap does not accept unsolicited resumes from recruitment agencies and search firms. Please do not email or send unsolicited resumes to any Nearmap employee, location or address. Nearmap is not responsible for any fees related to unsolicited resumes.