LinkedIn Live · Recordings

Livestreams

Long-form, unscripted sessions on AI in agriculture, hosted live on LinkedIn by Dr. Maryna Kuzmenko and kept here in full. Each one is aimed at the people who have to act on the answer — growers, agronomists, researchers, and the teams building the tools they use.

6 sessions 6 hr 22 min of recordings Full channel on YouTube
  1. Understanding Agricultural AI

    Prompt Engineering in Agriculture

    Premiered · 1 hr 11 min

    How to brief an AI assistant — Claude, ChatGPT, Gemini, Grok — like a colleague rather than an unpredictable vending machine. Covers the parts of a clear prompt, how to strip ambiguity out of farm data, grading and re-running a prompt until it holds, and why an agronomic decision still needs a human in the loop before anyone acts on it.

    • Prompt engineering
    • AI assistants
    • Office & field workflows
    • Human in the loop
    Watch on YouTube
  2. LinkedIn Live

    Seed Quality Control with AI

    Premiered · 59 min

    Computer vision applied to visual seed QC: counting, sizing, and flagging visible defects, colour differences and objects that should not be in the sample — demonstrated on chickpeas with Petiole Pro. Includes a warning worth the hour on its own: a model trained on soybean may transfer to chickpea, but its accuracy has to be re-checked against ground truth before you trust it.

    • Seed quality
    • Seed counting & sizing
    • Transfer learning
    • Petiole Pro
    Watch on YouTube
  3. Understanding Agricultural AI

    Collecting Agricultural Data for AI

    Premiered · 1 hr 3 min

    Reliable agricultural AI is decided long before the model sees an image. This session walks through what goes wrong in the field — shifting light, shadows, cut-off or occluded objects, a dirty lens, two phones processing the same scene differently, missing metadata — and the practical habits that make a visual dataset consistent enough to measure and compare.

    • Data collection
    • Image quality
    • Calibration
    • Metadata
    Watch on YouTube
  4. LinkedIn Live

    Hyperspectral Imaging in Agriculture

    Premiered · 57 min

    What hundreds of spectral bands reveal that a smartphone camera cannot. Compares RGB, multispectral and hyperspectral imaging, explains the hyperspectral data cube, and works through a study of 2,400 snap bean seeds used for protein prediction — alongside the parts rarely mentioned: computing load, equipment cost, and why lab measurements still do the validating.

    • Hyperspectral imaging
    • Seed quality
    • Protein prediction
    • Machine learning
    Watch on YouTube
  5. LinkedIn Live

    Multispectral Imaging & NDVI for Crops

    Premiered · 52 min

    The opening episode of the multispectral series, recorded as a Cereals 2026 “after party”. What NDVI actually represents, how it is built from red and near-infrared reflectance, and what an NDVI map can and cannot tell you — with the argument that an index is a starting point for investigation, never a diagnosis.

    • Multispectral imaging
    • NDVI
    • Precision agriculture
    • Plant health
    Watch on YouTube
  6. LinkedIn Live

    Sovereign AI LinkedIn Live

    Premiered · 1 hr 19 min

    What happens when a farm or agribusiness comes to depend on an AI platform it does not control. Separates raw data, derived data, trained models and the decision layer above them, then asks who owns each one when you leave a provider — plus the vendor lock-in, retention and re-use questions worth settling before you sign anything.

    • Sovereign AI
    • Data ownership
    • Vendor lock-in
    • Edge AI
    Watch on YouTube