Vaishnavi retweeted
Charts are now computable. Unsiloed can now turn time series plots into structured numbers that your AI agents can analyze, compare and calculate with.
VLMs can capture a chart’s structure and trend while getting the actual values wrong or reducing a dense series to a handful of points.
Unsiloed combines specialized agents with deterministic computer vision to extract detailed, structured data from complex charts, including 10 time series spanning ~250 trading days.
On a dataset built around chart patterns (that we have open sourced) we’ve seen pipelines struggle with:
• Lowest MAE among 7 pipelines: 6.56% of the axis range.
• Highest output density: 105,989 extracted coordinates.
Both matter since accurate points can still leave out most of a curve and more points alone don’t guarantee accuracy.
Already live with teams working on high-stakes use cases in finance with GA coming shortly. Working with time series plots or complex charts? Reach out to us.
Article link in comments.
Vaishnavi retweeted
Sharing my recent experience with interview process of an Indian IT company.
Because
Candidate experience matters.
My profile was shortlisted by Bootlabs for a AI Engineer role and an interview was scheduled.
I showed up for the interview, but it was cancelled because the interview panel was unavailable. I was told by the HR Manager that the interview would be rescheduled the following week due to panel availability.
I waited for a week and followed up yesterday.
Instead of a rescheduled interview, I received a generic rejection email saying they would not be moving forward with my application.
I have absolutely no issue with being rejected. Companies have every right to decide whom they want to hire.
What I find unprofessional is cancelling a scheduled interview because of an internal issue, asking the candidate to wait for a rescheduled interview, and then rejecting the candidate without even conducting the interview or explaining what changed.
This opportunity was important to me because I wanted to experience an actual AI Engineer interview, understand what I need to improve, and learn from the process , even if the final outcome was a rejection.
Candidates also invest their time, preparation and effort in these processes. That deserves the same level of professionalism from the hiring side.
Sharing this experience so other candidates are aware of how my hiring process with Bootlabs was handled.
Rejection is completely okay. Lack of transparency and communication is not.
Vaishnavi retweeted
In The Matrix, it becomes obvious it’s a simulation when the black cat appears twice.
World models fail similarly: visually the generated simulation looks realistic, but the underlying distribution is wrong.
Introducing CaliBench, a measurement of the randomness of reality!
Today, we're introducing CaliBench, evaluating whether video world models reproduce the true randomness of our universe.
Roll a dice or pick a card, the outcomes produced by a world model should match reality. Find out if do!
odyssey.ml/introducing-calib…
Vaishnavi retweeted
Nice work by the @xAI team on Grokipedia.com!
The goal here is to create an open source, comprehensive collection of all knowledge.
Then place copies of that etched in a stable oxide in orbit, the Moon and Mars to preserve it for the future.
Foundation.








