TargetSustainability data you can defend.
    Case Study · Agritech · Geospatial & ESG Analytics

    We make carbon measurable, mappable, and real.

    A carbon-monitoring platform needed to turn millions of land parcels and time-series measurements into decisions and reports stakeholders could trust. We designed, built, and now operate the interactive geospatial analytics platform behind it — from raw boundaries to sub-five-second insight.

    SHIPPED · OPERATING · IN PRODUCTION
    N 42°17' CRS · EPSG:4326 PARCEL LAYER · 2.4M POLYGONS
    Fig. 1 — Parcel layerCarbon attributes on a graticule
    Map-based geospatial analyticsTime-aware EAV data modelUnified transactional + analytical DBAurora Serverless v2Multi-tenant account isolationSub-five-second aggregationsNEE charts, monthly and annualInfrastructure as code
    The strategic outcome

    Trustworthy insight across huge geospatial datasets — in one unified system.

    ESG, carbon, and land-management platforms live or die on whether their numbers hold up. The value we delivered is not a dashboard; it is a foundation stakeholders can audit and defend.

    < 5s

    Trustworthy insight at scale

    Complex aggregations across millions of land parcels resolve in under five seconds — fast enough that analysis becomes a conversation, not an overnight batch job.

    One

    Unified system, no drift

    Transactional writes and heavy analytical reads share a single database. No separate warehouse, no sync latency, no divergence between what you record and what you report.

    Multi-tenant SaaS by design

    Strict account-level isolation and automated onboarding. Every tenant sees only their own estates, farms, and management units — reproducible and secure.

    Audit

    A defensible history

    A time-aware model records every land-use change, so any parcel can be queried exactly as it stood on any date. Stakeholders get numbers they can trace and defend.

    Signal, not noise

    Net ecosystem exchange, tracked month over month.

    Every parcel carries its own time-series of NEE measurements. The platform renders them as monthly and annual trends against a fitted baseline — so a change in practice shows up as a change in the curve, and a carbon claim can be traced back to the measurement behind it.

    < 5sComplex aggregation query time
    1Unified database, two workloads
    Tenants, isolated by account
    0Servers to manage
    +0 JanAprAugDec NET ECOSYSTEM EXCHANGE · gC / m² / mo
    How it is built

    The engineering that makes geospatial analytics fast — and defensible.

    Four decisions carry the platform: index on the ones that matter, keep the jargon purposeful.

    METHOD 01

    Interactive map-based analytics

    A full ArcGIS visualization layer for plotting and managing parcel polygons directly on the map — draw boundaries, filter by attribute, group by tag, and read carbon metrics in place instead of scrolling spreadsheets.

    Geospatial · Polygon plotting · In-map filtering
    METHOD 02

    Time-aware EAV data model

    An Entity-Attribute-Value model captures the full history of every polygon — attributes, practices, and land-use change over time. Land-use change is a first-class query, not an afterthought.

    EAV pattern · Historical tracking · Versioned
    METHOD 03

    One database, two workloads

    A single Aurora Serverless v2 (PostgreSQL) foundation serves transactional queries and analytical aggregations alike — accessed over the RDS Data API with no connection pooling to manage.

    Aurora Serverless · Unified · Zero-sync
    METHOD 04

    Serverless, isolated, reproducible

    Chalice, Lambda, API Gateway, S3, CloudFront, and Cognito — pay-per-use and automatically scaling. Cognito, JWT, and CloudFormation-defined infrastructure make every environment reproducible and secure.

    Cognito · JWT · Infrastructure as code
    The pipeline

    Polygons in. Decisions out.

    Raw boundaries and time-series measurements flow into a single serverless core, then out to an interactive, multi-tenant map dashboard — no separate transactional and analytical stacks to keep in sync.

    Parcels & polygons SOURCE 01 Time-series measures SOURCE 02 UNIFIED SERVERLESS CORE Aurora Serverless v2 TXN + ANALYTICS Chalice λ Lambda REST · JWT AUTH EAV TIME-AWARE MODEL RDS Data NO CONNECTION POOLS MULTI-TENANT DASHBOARD Filter · group · export · audit RAW BOUNDARIES + MEASURES ONE DATABASE · ONE SOURCE OF TRUTH SUB-5-SECOND INSIGHT
    Fig. 2 — Geospatial analytics pipeline · sources → unified serverless core → map dashboard
    PARCEL #A-1042 2023 2024 2025 tillage = conventional cover_crop = none NEE = +1.8 tillage = reduced cover_crop = rye NEE = −0.4 tillage = no-till cover_crop = rye + clover NEE = −2.1 EACH ROW = THE PARCEL AS IT WAS ON THAT DATE · NOT A FLATTENED SNAPSHOT
    Fig. 3 — Time-aware data model · one parcel, versioned across time (EAV)
    Spec sheet · how we build

    The stack, stated plainly.

    S-01
    Unified data architecture
    Transactional and analytical workloads on one Aurora Serverless foundation — no separate warehouse, no sync jobs, no divergence between what you write and what you analyze.
    S-02
    Serverless everywhere
    Chalice, Lambda, API Gateway, S3, CloudFront, and Cognito — pay-per-use, automatic scaling, and near-zero operational overhead from day one.
    S-03
    Time-aware by design
    An EAV model captures historical polygon attributes and management practices, so land-use change is a first-class query, not an afterthought.
    S-04
    Isolation you can trust
    Account-level data separation enforced across the stack, with automated onboarding and JWT-scoped access — multi-tenancy engineered in, not bolted on.
    S-05
    Infrastructure as code
    Every environment parameterized in CloudFormation and reproducible — reviewed like software, deployed like software, connection-pool-free via the RDS Data API.
    N 42°17' CRS · EPSG:4326 PARCEL LAYER · 2.4M POLYGONS
    Final assembly

    Bring us the geospatial problem that sounds too tangled to unify.

    Carbon data is complex, spatial, and time-bound — and so is our engineering. We have already shipped the trustworthy, auditable, real-time platform this space demands. Let's build yours.