Project overview

• Objective: Monitor the health and activity of a beehive using a network of sensors and AI to detect stress, illness or abnormal events.
• Scope: sensor data acquisition (environmental sensors + audio), signal/time-series analysis, event classification with AI, and a beekeeper-facing dashboard with alerts, history and recommendations.
• Reference: SEE26-IOT-01

Main responsibilities / Tasks

• Design and implement the IoT data pipeline: sensor integration, data logging, timestamping and communication (edge-to-cloud).
• Collect, preprocess and label sensor data (audio recordings, weight, temperature, humidity, vibration, etc.) and build datasets for supervised and unsupervised learning.
• Develop and train AI models for event classification and anomaly detection (audio classification, time-series models, and hybrid approaches).
• Integrate models into a dashboard: realtime/near-realtime alerts, historical visualizations and automated recommendations for the beekeeper.
• Evaluate models in-field and iterate using performance metrics and feedback from beekeepers.

Technical approach & tools (suggested)

• Sensors & IoT: microphones for hive audio, temperature/humidity sensors, weight scales, vibration sensors; data ingestion via MQTT/HTTP and local buffering on edge devices.
• Signal processing & features: audio preprocessing (spectrograms, MFCCs), time-series feature extraction, noise reduction and segmentation of hive activity periods.
• Machine learning: CNNs/RNNs/transformer-based audio classifiers, time-series models (LSTM, temporal convolution, Prophet/ARIMA for trends), and anomaly detection methods (autoencoders, isolation forests).
• Deployment & dashboard: containerized services (Docker), model serving (edge inference vs cloud), web dashboard (React/Vue + plotting libraries), alerting (email/push/webhooks).

Deliverables

• Working data ingestion pipeline and documented datasets collected from one or more hives.
• Trained and evaluated AI models for event classification and anomaly detection, with performance metrics and test results.
• A beekeeper dashboard with realtime alerts, historical views and actionable recommendations.
• Full code repository, deployment instructions, user documentation and a final project report describing methods, experiments and recommended next steps.

Candidate profile & skills

• Required: strong Python skills, experience with machine learning for audio/signal processing or time-series analysis, familiarity with data collection from sensors/IoT devices.
• Desired: experience with PyTorch or TensorFlow, edge deployment, web dashboard development and working knowledge of MQTT/REST APIs.

How to apply

• Apply online: https://lnkd.in/g3V53uMH
• Email subject (if applying by email): "Application - SEE26-IOT-01 AI-based Beekeeper’s Connected Hive PFE"