Abhinav Sharma
All work

Turning a vehicle appraisal from an afternoon of tab-juggling into four steps.

ClientAuto Appraiser Assistant
RoleFull-stack engineer
Year2026
Duration5 months
TypeScriptReactNodePostgreSQLVIN decode APIPDF generation
4Steps from VIN to finished report
6Appraisal tools in one workflow
FreeTo start — no card required

The problem

A vehicle valuation has to survive being argued with. A claimant disputes the number, and sometimes a lawyer does — so the figure needs comparables behind it, sourced and dated. Adjusters were assembling that by hand across separate places: a VIN decoder for specification, marketplace listings for asking prices, auction and salvage data for total-loss cases, then a word processor to staple it together. Slow, and worse, the evidence trail lived in whichever tabs happened to be open when the report was written.

The approach

One workflow from VIN to finished document. A VIN decode establishes the specification, comparables are pulled from live market listings alongside salvage and auction records, and every figure that lands in the report keeps a reference back to the listing it came from. The PDF is generated from that same evidence rather than transcribed into it, so the number in the claim file and the listings supporting it cannot drift apart.

The outcome

An appraisal that used to mean juggling several sources and a word processor is four steps, and the report a claimant receives carries the comparables it was built from.

What mattered

Evidence travels with the number

Every figure in the report keeps a reference to the listing behind it, so a disputed valuation can be defended rather than re-researched.

Salvage and auction data included

Total-loss cases need comparables that ordinary marketplace listings do not provide, so that data sits in the same search.

The report is generated, not written

Producing the PDF from the same records used to reach the valuation removes the gap between what was researched and what was filed.