ChronoShelf: A prescriptive inventory engine for agrivendors

We combine Arrhenius decay physics with Spark-driven optimisation to predict shelf-life and align stock with shifting local demand. Vendors can use this app to plan their warehouse stock orders better

Comment

Inspiration-

Cold storage owners across India lose ₹1 Lakh+ daily to crop spoilage — not from lack of data, but from lack of predictive tools. We wanted to apply thermodynamics, not guesswork, to this problem.

What it does-

ChronoShelf predicts when each crop will spoil using the "Arrhenius decay equation" — the same kinetics model used in pharmaceutical shelf-life studies. It ingests real-time warehouse temperature logs, simulates decay rate per crop, and fires "Sell-Now alerts" before spoilage occurs.

How we built it-

  • "Delta Lake" stores a time-versioned thermal history for each crop batch
  • "Apache Spark" runs the Arrhenius integration across all vendors in parallel
  • "Plotly dashboard" surfaces a heatmap of decay state + actionable restock recommendations
  • Crop-specific activation energy (Eₐ) constants for Potato, Onion, and Tomato power the physics engine

Challenges-

Converting continuous temperature logs into integrated decay — accounting for temperature fluctuations rather than assuming steady-state — required careful numerical integration in Spark.

What we learned-

Delta Lake's time-travel capability is a natural fit for tracking physical state evolution. Most teams use it as a flat database; we used it as a thermodynamic ledger.

What's next-

Expand crop constants library, add humidity as a second decay variable, and deploy as a mobile-first vendor app.

Built With

Share this project:

Updates