Darell
21 posts
Daqstra works with the equipment teams already have, including NI/LabVIEW, PLCs, custom DAQ systems, sensors, controllers, and vendor hardware.
We’re already working with advanced hardware teams across aerospace, energy, and drones, including active aerospace deployments and an energy pilot targeting $100K+ in annual savings.
If you lead test, propulsion, integration, validation, or hardware engineering, we’d love to talk.
Feel free to DM me or schedule a demo on daqstra.com
English
Before each test, we connected sensors, custom electronics, NI hardware, PLCs, and controllers that were never designed to work together.
After each run, data was scattered across machines, file formats, scripts, and people’s heads. Before we could learn from one test, we often had to reconstruct what happened.
As we talked with teams building rockets, energy systems, robots, and drones, we kept hearing the same story:
Their hardware was advancing faster than the infrastructure used to test it.
So we built Daqstra.
English
Test infrastructure shouldn’t bottleneck great engineering.
Today, we’re launching @daqstra with @ycombinator.
Daqstra turns complex test infrastructure into an AI-native operating layer.
We’re Anchit and Darell, the founders of Daqstra.
We first met while testing liquid rocket engines in the Mojave Desert, where we realized much of the friction wasn’t the engines—it was the infrastructure around them.
English
Daqstra works with the equipment teams already have, including NI/LabVIEW, PLCs, custom DAQ systems, sensors, controllers, and vendor hardware.
We’re already working with advanced hardware teams across aerospace, energy, and drones, including active aerospace deployments and an energy pilot targeting $100K+ in annual savings.
If you lead test, propulsion, integration, validation, or hardware engineering, DM me or schedule a demo on our daqstra.com
English
Before each test, we connected sensors, custom electronics, NI hardware, PLCs, and controllers that were never designed to work together.
After each run, data was scattered across machines, file formats, scripts, and people’s heads. Before we could learn from one test, we often had to reconstruct what happened.
As we talked with teams building rockets, energy systems, robots, and drones, we kept hearing the same story:
Their hardware was advancing faster than the infrastructure used to test it.
So we built Daqstra.
English
Daqstra makes your infrastructure a programmable software platform. It works with the equipment teams already have—including NI/LabVIEW, PLCs, custom DAQ systems, sensors, controllers, and vendor hardware.
We’re already working with advanced hardware teams across aerospace, energy, and drones, including active aerospace deployments and an oil/energy pilot targeting $100K+ in annual savings.
If you lead test, propulsion, integration, validation, or hardware engineering, we’d love to talk.
Go to daqstra.com to schedule a demo or send me a DM.
English
Before each test, we connected sensors, custom electronics, NI hardware, PLCs, and controllers that were never designed to work together.
After each run, data was scattered across machines, file formats, scripts, and people’s heads. Before we could learn from one test, we often had to reconstruct what had happened.
As we talked with teams building rockets, energy systems, robots, and drones, we kept hearing the same story:
Their hardware was advancing faster than the infrastructure used to test it.
So we built Daqstra.
English
Introducing Nyx, the world’s frontier hacker of AI agents.
On AgentHarm, Nyx achieved a 78% attack success rate - compared with 67% for GPT-5.6 Sol.
Here’s more about how it works 🧵
English
Enterprise agent adoption today is concentrated in short, single-turn like routing, extraction, classification, tool calls. These workflows will soon run at trillion-token scale. These are exactly where a post-trained small model beats the frontier on cost and latency.
We believe that post-training techniques and environment infra will diffuse, and the value value will concentrate in automating the process: turning your data and traces into datasets, environments, rewards, evals, and training loops with no human in the middle.
By creating inexpensive, efficient training infrastructure, iterative post-training is now also viable. Flash makes the full loop agent-completable.
Freesolo@freesoloco
Not every AI interaction is best served with a large frontier model. There is a long tail of trillion-token use cases, from tagging to search, best served by a sub-10b-parameter model that runs in milliseconds and costs many orders of magnitude less than the frontier. We built Freesolo Flash to make training loops like SFT and RL easy and end-to-end completable through your coding agent.
English
We thought building a cryogenic rocket engine would be the hard part.
It wasn’t.
The real bottleneck was everything around the test: fragmented hardware, brittle scripts, manual procedures, redlines, and hours spent digging through logs.
So we built Daqstra: an AI-native test orchestration platform that brings hardware, live data, procedures, automation, and analysis into one place.
Today, we’re excited to share that Daqstra is part of Y Combinator S26.
If your team runs complex physical tests, we’d love to talk.
English
Can’t wait to show what we’ve been building. Reach out if you’re interested!
Daqstra@daqstra
We’re joining @ycombinator for the summer 2026 batch!
English
@PatrickFinley_ Great news Patrick. Incredible for collegiate rocketry!
English
We just partnered with the Musk Foundation to give out $175k in grants to advance the frontier of collegiate rocketry.
This is in addition to the $390k we’re already giving out to collegiate teams building propulsive self-landing rockets.
When we first started, people laughed and told us what we were attempting would be impossible.
2 years later, several student teams have achieved both TVC and throttled hotfire with liquid engines, and they are making fast progress on their hoppers. We’re already seeing hover attempts at a collegiate level!
These kids are doing this on 0.001% of a typical rocket budget, dedicating 80 hour weeks without any pay to build some of the most capable rockets outside of industry.
The reason why so many people fall in love with space and rocketry is because it is proof that humanity is capable of the impossible.
We want to continue to advance that mission.
Back to work.
English
How a college rocketry team professionally tested a YC startup’s heat shield.
Daqstra was the orchestration layer behind the test.
daqstra.medium.com/how-a-college-…
English
Darell retweetledi
English
Darell retweetledi
skills launch party at shack15
agent skills are taking over the tech world right now
my favorite from the event is from @mintlify where you can just add any doc page as a skill directly
npx skills add https-your-docs-domain-dot-com
English