LOGISTICS · AI · Real-time Data · Optimization

    Intelligence on every mile

    Dynamic routing for a 5,000-vehicle national fleet.

    NATIONAL LOGISTICS COMPANY · 200 CITIES // 22 WEEKS // BOLT GROUP

    $18MFuel saved annually
    Built with
    Apache Kafka
    Python
    PyTorch
    PostgreSQL
    Redis
    Kubernetes
    Mapbox
    AWS
    The situation

    A $120 million annual fuel budget, 20% of it wasted on suboptimal routing. One in five deliveries missed its window. Routes were planned once daily and didn't adapt to traffic, weather, or closures. Fleet utilization sat at 64%. Every problem was a data problem in disguise.

    Why it mattered
    $120MAnnual fuel budget
    1 in 5Deliveries missed window
    DailyRoutes planned once, never adapted
    64%Fleet utilization baseline
    client
    National Logistics Company · 200 cities
    duration
    22 weeks
    services
    AI/ML, Cloud, Data Engineering
    sector
    Logistics & Transportation
    The approach

    Four moves, built to compound.

    1. 01

      Real-time data fusion

      Streaming platform ingests live traffic, weather, GPS telemetry, and delivery schedules into a single operational picture.

    2. 02

      Optimization engine

      Constraint-based model using genetic algorithms and reinforcement learning.

    3. 03

      Dynamic re-routing

      Continuous monitoring detects disruptions — accidents, weather, closures — and recalculates affected routes with driver notifications.

    4. 04

      Fleet command center

      Real-time tracking, performance metrics, fuel analytics, and predictive ETA across the entire national fleet.

    Delivery discipline
    • Principal-ledSenior engineers and architects on the keyboard from week one. No staffing pyramid, no handoffs to juniors mid-engagement.
    • Production-firstEvery artefact ships behind a load balancer. We measure success in throughput, latency, and uptime — not slides.
    • Outcome-boundScope, milestones, and acceptance criteria locked at kickoff. We commit to the metric that moves your P&L.
    Technology
    1. 01Apache Kafka
    2. 02Python
    3. 03PyTorch
    4. 04PostgreSQL
    5. 05Redis
    6. 06Kubernetes
    7. 07Mapbox
    8. 08AWS
    22 WEEKS · AI/ML · CLOUD · DATA ENGINEERING · SHIPPED
    The outcome

    What shipped. In production.

    0%
    Fuel saved
    $18M annually
    0%
    Faster delivery
    On-time performance
    0%
    Fleet utilization
    Up from 64%
    0
    Vehicles
    Across 200 cities
    Real-time
    Re-routing
    Continuous adaptation
    22wk
    Delivery
    End-to-end deployment
    The insight
    "Logistics margin compounds with intelligence. Dynamic routing pays back in the first fuel cycle — and every cycle after."
    BOLT GROUP // PARTNER'S NOTE // BOLT GROUP // PARTNER'S NOTE // LOGISTICS
    How we engage

    Three shapes. One delivery standard.

    A / Fixed-scope build

    Defined outcome. Fixed price.

    We scope, build, and ship a discrete production system end-to-end. Best when the brief is sharp and the deadline is real.

    B / Retained capacity

    Embedded pod. Monthly cadence.

    A senior pod plugs into your team for a multi-quarter horizon. Roadmap evolves; the engineers stay constant.

    C / Advisory & audit

    Senior eyes. Surgical scope.

    Architecture review, model audit, or technical due diligence. Days to weeks. Written recommendations, not slideware.

    Start the conversation

    Let's build what's next.

    No SDR queue. No discovery deck. Pick a slot or send a note — Dean replies within four business hours.