DraftMark uses a model to write ship survey code, checked against a printed booklet

·15 views

A developer built DraftMark for the Meta AI Developer Hackathon to time-average a ship's moving waterline from video and have a model generate per-ship draft survey programs. Rather than trusting the model's output, the author tested the generated program against 22 worked loading conditions printed in a Ro-Ro ship's stability booklet, requiring agreement within 0.1 percent. A verbatim implementation of the booklet's procedure matched only 14 of 22 rows, because it omitted a second trim correction and contained unstated conventions. After two regenerations with feedback on failing rows, the model's program reproduced all 22 conditions. The finished survey was also checked on one in-sample row, landing about 0.01 percent from the printed value.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

FINDR AI Open-Source Tool Targets GPS Spoofing in Parcel Delivery Tracking

FINDR AI is an open-source system built to detect suspicious location manipulation in e-commerce parcel delivery, where GPS coordinates can be misleading. It flags GPS spoofing, physically impossible teleportation jumps, and gradual location drift, then assigns risk scores to suspicious telemetry. The system adds investigation workflows, human approval for critical rescue actions, and hash-chain audit logging for traceable decisions. Built in Python with Streamlit, FastAPI, MongoDB Atlas, and SQLite, it was submitted to the Hacktoberfest Open-Source AI Challenge and reports 105 passing regression tests.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Debugging Malformed JSON From an Email Send API After Payment

The article advises validating the exact request payload against the provider's documented schema before sending, rather than assuming the response is at fault. It recommends checking the HTTP status before parsing the body, since an HTML gateway error or empty response can cause a JSON decoding failure. Retries should use a stable idempotency key and a policy keyed to status codes, to avoid duplicate sends and redundant payloads. The author also proposes splitting retention, keeping operational metadata and a payload hash longer while shortening retention of sensitive receipt content. The overall guidance is to log request bytes, status, content type, and a size-limited raw body separately from parser errors.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Measuring a Server: Latency Knees, Misleading Percentiles, and Little's Law

The author built a deliberately simple single-worker server to study how latency behaves under load. Ramping request rates revealed a 'knee' where tail latency rises sharply well before the server is fully utilised. At about 70% busy, the 99th percentile latency had already more than doubled. The author also argues that averages hide slow outliers, so percentiles better reflect what users actually experience. A further experiment showed that fan-out amplifies this effect, since a page is only as fast as its slowest dependency.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Developer Struggles With Calculator App Logic, Plans to Resume Work Tuesday

A developer posting on DEV Community says they are stuck on the logic of a calculator app they are building. They plan to continue the project on Tuesday because they have other commitments on Monday. The developer had expected the task to be simple but found it more difficult than anticipated. They recalled previously building a basic calculator webpage with AI assistance in about half an hour, and reflected that they have not retained the earlier learning.

0 comments Read more at DEV Community