Tree Species Interpretation Specialist

Flash Forest

9h•Remote

About The Position

Flash Forest is seeking a Tree Species Interpretation Specialist for a fully remote, hourly contract role. This position supports their automated reforestation efforts by identifying tree species from drone imagery. The specialist will work closely with the machine learning and Forest Intelligence team, providing labels that train models to recognize species at scale. The accuracy of the labels directly impacts the company's species classification for clients. The role involves interpreting individual tree images derived from RGB orthomosaics, supported by LiDAR data, multispectral bands, and ground truth maps.

Requirements

Several years of field forestry or silviculture experience in boreal forests and/or Interior BC (e.g., silviculture and free-growing surveys, regeneration surveys, cruising, or planting quality inspection).
Strong tree species identification skills, including confident separation of similar species (e.g., black vs. white spruce, aspen vs. poplar, birches, maples).
Familiarity with Ontario boreal and Great Lakes-St. Lawrence species.
Detail-oriented and consistent.
Ability to work independently, manage own hours, and meet batch deadlines.

Nice To Haves

Aerial photo interpretation experience (e.g., VRI photo interpretation or forest cover mapping).
Experience working with multispectral imagery or vegetation indices.
Familiarity with LiDAR products such as canopy height models.
Previous experience labelling data for machine learning.
RPF or RFT designation, or equivalent.

Responsibilities

Review individual tree images and point cloud data to assign or validate species labels, focusing on major species in central and northern Ontario and BC (Conifers: Jack Pine, Black Spruce, White Spruce; Hardwoods: Trembling Aspen, Balsam Poplar, White Birch, Yellow Birch, Sugar Maple, Red Maple, Oak).
Interpret RGB imagery using crown shape, texture, colour, branching pattern, shadow, and stand context to identify species.
Use LiDAR-derived height and crown form alongside imagery to support difficult calls.
Cross-check labels against ground truth plot data and field photos, and flag mismatches or likely errors.
Flag ambiguous, overlapping, dead, damaged, or non-tree crowns, and record a confidence level for uncertain calls.
Help build and refine a species interpretation key and labelling guidelines.
Interpret multispectral data (e.g., red edge and near-infrared bands, vegetation indices) to help separate species and species groups that look similar in RGB (bonus).
Track hours and progress, and document labelling decisions for consistent application.