SyntheticRows generates realistic synthetic tabular data from your own CSV. Unlike most tools that hide behind a single quality score, it explains data quality with three transparent metrics (distinguishability, statistical similarity, and coverage) plus a real ML-readiness test that trains on synthetic data and evaluates on real data. It also detects class imbalance, explains data issues before generation, supports hybrid mode for free-text columns, and is free with no signup.

Hey Product Hunt 👋 I'm Dhruv, a final-year AI student in India. I built SyntheticRows because every synthetic-data tool I tried for coursework and side projects gave me a quality score I had no reason to trust — often suspiciously high, with zero explanation. So I built the version I wanted: one that profiles your data, explains every issue it finds before touching anything, generates synthetic rows, and then honestly tells you whether the result is good enough to train a model on — including saying "inconclusive" when it can't tell. It's live, free, and I built and shipped the whole thing solo (FastAPI + SDV backend, Next.js frontend). Would love your feedback, bug reports, and brutal honesty — especially from anyone who's been burned by an inflated synthetic-data score before.

The three-metric breakdown instead of a single score is genuinely refreshing, and the ML-readiness check actually gave me useful signal on my dataset. Free with no signup made trying it painless.

SyntheticRows

Turn a small CSV into a bigger dataset — scored honestly

Turn a small CSV into a bigger dataset — scored honestly