KRAstra

The problem

Employee performance data lived in the TMS, in BI reports and in local spreadsheets. It was downloaded, copied and re-processed by hand into individual reports. Targets came from the individual’s own feedback, not from the potential of their area. Nobody could send a person their monthly score.

What I built

A scorecard and performance-review platform that doubles as a management dashboard. Role-based KPIs across zone, region, area and branch on the Indian financial year. Benchmark scoring that a person can understand: one number per KPI, score equals actual over benchmark. Effective-dated role assignments so a role change never rewrites last quarter. Weighted composites for people who hold more than one role. Scores computed once when HR publishes, so dashboards are fast and free. An audit trail of every admin change. CSV export. Archive, never delete.

Shipped, on Firebase, in September 2026.

What happened

Nobody used it. Not because the software failed, but because the baseline data it needed was never being tracked in the first place. A scorecard with no actuals is a very well-designed empty page.

What I learned

• Fix data before building software, not just before buying AI. I had the principle; I applied it to AI and not to my own product.
• Ask who will feed it, every month, before writing a line. The answer here was “nobody yet”.
• Ship the data pipeline first and the dashboard second, even when the dashboard is the part people get excited about.

Why it is on this site

Because every portfolio is a list of wins, and the thing that changed how I work was a loss. The KPI standardisation work that preceded it (1,783 rows into 57 KPIs) still stands; the platform is ready when the data is.