I've been building Calibra, an open-source dataset observability and coreset optimization tool for robot learning.
It checks things like timestamps, duplicate/frozen frames, blurry cameras, jerky motion, action-state divergence, quality, and behavioral coverage.
The interesting part is the optimization side: across several public robotics datasets, we've found that quality-aware selection can preserve model performance while using substantially less training data than the full dataset or random selection.
Open Source project Which Let's you Understand your Codebase in Detail with AST Tree-Sitter Parsing with a Bring Your Own key (BYOK) Architecture Where a User Can Bring their Own API Key for LLM Calls so that you codebase remains Private to you, Along with that the Product Creates a Dependency Graph and also Make a Detailed Report for the Codebase, The Dependency Graph Looks Exactly Like the Graphs Shown in the Neo4j Graph Database.
Link :- https://gitdecode.app
Github :- https://github.com/Priyanshu-Deb... Connect Me On Linkedin :- https://www.linkedin.com/in/priy...
I've tried agent orchestrators like Conductor but I've never really stuck to it long term because I have no way of verifying the huge amount of work they've done with basically no validation. I always end up going to my one-agent Claude Code workflow and always feel like I'm missing out.
What kept happening to me:
3-5 agents fan out, each opens a PR
CIs pass, diffs look reasonable
I can't possibly click through every preview by morning
I merge based on the diff and CI signal
Something breaks in prod the next day because the agent did what I asked literally, but the feature doesn't REALLY work
I am building Anotee after watching client feedback get scattered between email, chat and shared review links.
The premise is simple: every note should stay attached to the exact timecode and video version it refers to. I am looking for ten editors or small production teams to try the workflow on a real project and tell me where it fails.
This is my product, not an independent recommendation. If you use Frame.io, email, WhatsApp, or another tool today: what breaks first when a third review round begins?
I'm a QA Lead who's worked with well-known brands such as Apple, IKEA, Thales, and Gap.
LinkedIn helped me land some of those opportunities because I was active on the platform with a long-term content plan. This means I took the time to craft a consistent tone of voice that sounded like me, defined the content pillars that guided my posts, and ensured that every time I posted, I brought in my own perspective, learnings, or ideas.
Two people told me the same feature surprised them. So I'm rebuilding it around that instead of guessing.
Quick update on FounderMind AI. Since the last launch thread, two separate people mentioned the same thing without prompting: the competitor-analysis lens surfaced something they hadn't actually found on their own one person a niche competitor they'd missed, another said the pricing suggestions felt grounded in real market context rather than a generic template.
Earlier this year I found myself job hunting again after 8+ years in product, most recently as a Product Owner in UK banking.
A job hunt gives you almost no feedback. You apply, nothing comes back, and you are left guessing whether the role was never a fit, the application was weak, or the ad was never real. With nothing to go on, everyone reaches for the same two moves: rewrite the CV, or apply to more jobs.
I did both, for months. Neither was the problem.
WorkstationAI is a workspace for running the whole hunt, and last week I shipped the part that closes the loop.
You no longer need to start with an existing MIDI file. ShevtoneAudio can now create music from scratch.
Generate original music automatically Create complete MIDI compositions Turn ideas into cinematic arrangements Great for cinematic, trailer, game & film music Download the MIDI and continue editing in your favorite DAW
From a blank project to an editable MIDI composition in minutes.
