News for hackers. Stuff that matters.

mattismegevand

Show HN: Repere – Local-first SQL data explorer, no uploads needed

https://repere.ai

Repere lets you drop CSV/JSON/Parquet/XLSX files into your browser and immediately query them with full DuckDB SQL. Nothing gets uploaded, everything runs locally via DuckDB WASM. Unlike Excel and Google Sheet it can handles millions of rows easily.Features: - Filter, sort, join, pivot across multiple files Every transformation becomes a node in a visual pipeline (DAG) - Full undo/redo,...

( Read More... | no comments | on HN )
Re:Ask HN: What Are You Working On? (January 2026) on Sunday January 11, 2026 @04:58PM (#46580686)
I built an in-browser data explorer for datasets too large for spreadsheets but too small to justify spinning up SQL. It runs entirely client-side using DuckDB-WASM, no server uploads accounts or anything.

The core idea is a visual DAG where each transformation (filter, join, aggregate, pivot) creates a view node. Nothing materializes until you need it, DuckDB executes the full chain on demand so you can build deep pipelines without copying data at each step.

Input files can be CSV/Parquet/Excel (Excel might not work great). There's a SQL editor with schema-aware autocomplete, pivot tables with drill-down to underlying rows, and sessions can be exported as files or shareable URLs (the entire pipeline gets encoded in the hash).

Sharing can be granular and you can choose not to embed the files or if files are too big they will not be embedded and the user when opening the link will have to upload the files to restore the session.

The part I find most useful: you can replay pipelines on new data. Share your January analysis, and a colleague runs the same transformations on February's data with schema validation.

Privacy-first since files never leave your browser, it's a static website actually. I will open source soon, and make it probably MIT licensed.

Also it's a WIP and so it may be buggy (there's not even images on the homepage) https://repere.ai

Close to 100%, this problem seem to be easily solved by the model. I plan to do cross validation and also to explore the use of a smaller model.
The two big cluster are indeed photo and charts. However chart or meme are not all invertible, charts with lots of color can for example be degraded by inversion. Also some chart are already done for dark theme, for those images the model will understand that it doesn’t need to invert them.
Exactly, the transformation done for inference are located here: https://github.com/mattismegevand/invertornot/blob/main/api/...
Thank you! Very little resources are needed to run it, gwern is using the API and at first I was using a 6$ Droplet (exact same specs as the t2.micro) and it was running well.
Some graphs with lots of colors can be difficult for models to handle. In these cases, the model often chooses not to flip the graph to avoid making it hard to read, based on past problems with similar graphs. The model generally does a good job at deciding when to flip something. Although I haven't seen it flip photos, it's possible that there are examples out there.
Thank you, yes you can try it directly (https://invertornot.com/docs) or you can either upload an images or spam the button to get random images and see their prediction.

Show HN: Invertornot.com – API to enhance your images in dark-mode

https://invertornot.com

Hi HN, I built (https://invertornot.com) it's an API that can predict whether an image will look good/bad while inverted. This is particularly useful for images in dark-mode as you can now safely invert them.The conservative solution to adapt images for dark-mode consist in dimming the image, however there is a lot of images that can be inverted (graph for example). Using deep learning we can...

( Read More... | 38 comments | 74 points | on HN )