Current build
1. RunLedger
Import running activity files, curate datasets, and establish the trustworthy history layer the rest of the platform depends on.
ClubLab PreviewRunLedger active
Open running data platform
ClubLab is an OSS-first platform for turning raw activity data into a structured product: from ingestion and storage to modeling, analytics, and eventually simulation. It is designed as a compact, end-to-end environment for building and validating training and performance workflows on a modern OSS (free) data stack.
The current build, RunLedger, focuses on the foundation: reliable imports, dataset curation, and a clean analytics layer. This establishes the baseline needed for higher-order features like prediction, planning, and scenario testing.
This preview exposes the full shape early — ingestion, secure storage, orchestration, and modeling — as a single system, not a collection of tools.
Now, next, later
ClubLab is intentionally staged. RunLedger proves the ingestion and warehouse foundations first, then the same platform expands into richer analytics and training simulation.
Next layer
2. Training analytics
Model progress, coverage, recovery, and training trends on top of the same warehouse so operational data becomes coachable insight.
Future layer
3. Simulation + prediction
Use the curated data platform to test training scenarios, estimate outcomes, and demonstrate model-assisted planning workflows.
Stack
Opinionated, OSS-first building blocks for ingestion, modeling, analytics delivery, and future simulation work.
Next.js
Preview site, product UI, and future analyst-facing surfaces
TypeScript
Shared contracts across web, services, and data tooling
Postgres
Warehouse for raw runs, curated datasets, and analytical marts
Supabase
Auth, RLS, storage, and secure multi-user foundations
dbt
Transform runner history into analytics-ready models and metrics
FastAPI
Operational APIs for imports, enrichment, and future modeling workflows
Python
Parsing, enrichment, prediction experiments, and simulation prototypes
Prefect
Schedules, retries, and orchestration across imports and refresh jobs
Redis
Optional queueing, coordination, and rate limiting for worker flows
Metabase
Dashboards for training trends, coverage, and pipeline operations
Platform map
End-to-end view of how RunLedger fits inside the broader ClubLab platform and how the shared data model can later support analytics, prediction, and training simulations.
Experience
Runner / coach
Authenticated context and training intent
uses
ClubLab UI
Preview, workbench, and future analytics views
Control Plane
Supabase
Auth, RLS, and file storage
auth
APIs + workers
Imports, enrichment, and job coordination
schedules
Prefect
Schedules, retries, and backfills
api guardrails
Redis
Queues, locks, and worker guardrails
Supabase
Auth, RLS, and file storage
auth
APIs + workers
Imports, enrichment, and job coordination
api guardrails
Redis
Queues, locks, and worker guardrails
schedules
Prefect
Schedules, retries, and backfills
Data Plane
Compute Inputs
Wearables + files
FIT exports, manual uploads, external feeds
Modeling jobs
Python enrichment, dbt transforms, simulations
Source of Truth
Postgres warehouse
Raw runs, curated datasets, metrics, and training marts
Consumers
RunLedger
Dataset curation and import operations
Metabase
Training trends, quality checks, and ops dashboards
Prediction surfaces
Scenario analysis and simulation outputs
Compute Inputs
Wearables + files
FIT exports, manual uploads, external feeds
Modeling jobs
Python enrichment, dbt transforms, simulations
Source of Truth
Postgres warehouse
Raw runs, curated datasets, metrics, and training marts
Consumers
RunLedger
Dataset curation and import operations
Metabase
Training trends, quality checks, and ops dashboards
Prediction surfaces
Scenario analysis and simulation outputs
Today the flow starts with run ingestion and dataset curation; later it expands into dashboards, simulation services, and predictive training experiences.