This role is with Polybee, a startup supported by SGInnovate.
Background
Polybee is reducing uncertainty in fresh produce agriculture with autonomous pollination and yield forecasts using drones and AI. This can unlock over $20B+ in value to farmers through increased yields and reduced waste. Polybee’s autonomous drone pollination can improve yields by up to 20% while accurate forecasts improve profitability through efficiency in harvesting and informed pricing. Polybee’s product is already live and scaling up at the biggest growers in ANZ, and they're now making progress with multiple fresh produce giants in the US. Polybee is a Singapore and Melbourne-based startup backed by Temasek Foundation, SEEDS Capital, elev8 VC among others.
Job scope
What you’ll do at Polybee
- You will build and productionize deep-learning models that detect, segment and extract actionable insights from fruits and vegetables using imagery captured by off-the-shelf drones, achieving industry-leading accuracy on key KPIs.
- You will own the end-to-end data and model pipeline—from raw image ingestion and data pipelines through training, validation, deployment and automated retraining—ensuring client requests.
- You will optimize inference pipelines for both cloud and embedded edge hardware using runtimes such as TensorRT, ONNX Runtime or equivalent, meeting latency targets and resource budgets.
- You will implement monitoring, drift detection and automated feedback loops so that any degradation in model performance is identified and corrected.
- You will establish and enforce a quality-first code culture: unit/integration tests, code reviews and CI/CD (e.g. GitHub Actions or equivalent), so that 100% of changes roll out with automated validation and zero manual steps.
- You will partner with robotics engineers to refine drone flight patterns, camera settings and preprocessing workflows, improving input raw data completeness, quality, latency and cost.
- You will collaborate closely with operations teams and key customers to ensure field deployed solutions achieve promised results, incorporating direct feedback from initial deployments into roadmap decisions.
- You will design and manage the full MLOps lifecycle, from data validation and model versioning to CI/CD pipelines for models, shadow deployments, and rollback mechanisms.
- You will contribute to the creation of intellectual property and defensible innovation through novel algorithms, research publications, or patents that position Polybee as a category leader.
Here’s what we’re looking for
- 2+ years building and shipping computer vision and deep learning solutions in production, with demonstrable impact on business metrics (accuracy, latency, cost)
- Strong programming skills in Python and C++, with experience in computer vision/ML libraries (e.g., OpenCV, scikit-learn, pandas) and deep learning frameworks (e.g., PyTorch, TensorFlow or equivalent)
- Hands-on involvement in designing and maintaining ML pipelines with data/model versioning (e.g., MLflow, DVC, Prefect or equivalent) and containerized deployments (Docker, Kubernetes or equivalent)
- Proven ability to define, track and present KPIs via dashboards (Grafana, Prometheus or equivalent) and close feedback loops to sustain model performance
- Excellent communication skills, capable of translating complex technical concepts for non-experts and mentoring junior engineers
- Ability to work directly with operations and customers to deploy and refine AI powered systems in real-world field conditions
- Technical expertise in 3D reconstruction, stereovision, photogrammetry or structure-from-motion methods will be highly valued
- Experience creating IP—whether in the form of patent filings, conference papers, or proprietary research—will be highly valued
Level of relevant working experience
3 - 5 years
Required Skills
- Python
- Machine Learning
- Cloud Computing
- Computer Vision
- Deep Learning
Interested candidates may apply directly at: *************;keyword=polybee