SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World
Back to feed
Programming

Microsoft Offers Voluntary Exits to About 60,000 Eligible Employees

0
·2 views

Microsoft announced a Voluntary Exit Offer on October 26, 2026, covering eligible employees at level 70 and below in Canada, India, the United Kingdom, and the United States. Participation varies by organization, and the program includes financial support and career-planning resources. Reuters reported that roughly 60,000 employees are eligible, though the offer is voluntary and does not mean that many people have already been dismissed. Microsoft has not linked the departures to automation or AI, describing the move instead as adapting its workforce to changing needs. The eligible group also includes employees who are not developers.

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)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Mixture of Experts: How Sparse Routing Scales Model Size Without Matching Compute Costs

Dense transformers spend the same compute on every parameter for every token, so making them larger also makes them more expensive. Mixture of Experts (MoE) instead keeps a large pool of parameters but activates only a small subset for each token. A learned router scores the experts and selects the top few, so total model size can grow far faster than per-token cost. This design introduces a risk: the router can concentrate traffic on a few experts while others go unused, a problem typically countered with an auxiliary load balancing loss. The article explains the routing mechanics, why balancing matters, and when MoE may be the wrong architectural choice.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Recomputing Collective2 Calmar ratios on four strategy pages yields mismatches; profit factors mostly hold

A developer tried to reproduce the Calmar ratios and profit factors published on four Collective2 strategy pages using the figures each page shows. Profit factors for three of the four pages matched their recalculated values within about 3%, while one differed by roughly 15%. None of the four Calmar ratios matched the recomputed values, with the published figures running between 1.21 and 1.53 times higher. The author says the variation suggests a platform-side difference in how the metric is defined rather than deliberate misrepresentation, but could not confirm the cause without trade-level data or an equity curve. The check was limited to four pages from a single day, and the author notes it says nothing about future performance.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Preserving Notary Stamps and Signatures When Converting Scanned Legal Documents to Editable Word

Scanned legal documents mix editable typed text with graphic authenticators such as notary seals, watermarks, and handwritten signatures, which is hard to convert. Conventional tools often rasterize pages into images that cannot be edited, while standard OCR misreads curved stamp lettering and injects garbled characters into clauses. Some desktop tools use floating text boxes with white fills that hide underlying seals and watermarks, making the document legally unusable. The article proposes a dual-layer approach that separates a reflowable, fully editable paragraph layer from a graphic overlay layer holding stamps and signatures. Color-based filtering is used to segment colored ink and stamp strokes from black toner text, so authenticating marks can be preserved as transparent overlays.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Editing an approved draft invalidates its approval, a browser test shows

A developer building a local draft-review prototype found that an approval should attach to the specific reviewed text rather than a fixed row ID. The prototype's checks reset when text was edited, which Playwright assertions confirmed in Chromium, and downloaded CSV files were inspected to verify the correct content was exported. The author noted the prototype stores changes only in memory and that the tests did not cover other browsers, accessibility, or server-side enforcement. The author suggests binding approval to a content version and having the server reject exports when that version no longer matches, though this design was not built or tested. The prototype was shelved because a spreadsheet was a strong substitute and there was no buyer evidence.

0 comments Read more at DEV Community