Rushil Agarwal (@Rush26agarwal) profile picture

Founder at Human Archive (YC W26) | UC Berkeley MET (dropped out)

Joined February 2025
Photo by Rushil Agarwal (@Rush26agarwal)Photo by Rushil Agarwal (@Rush26agarwal)Photo by Rushil Agarwal (@Rush26agarwal)Photo by Rushil Agarwal (@Rush26agarwal)Photo by Rushil Agarwal (@Rush26agarwal)Photo by Rushil Agarwal (@Rush26agarwal)Photo by Rushil Agarwal (@Rush26agarwal)Photo by Rushil Agarwal (@Rush26agarwal)Photo by Rushil Agarwal (@Rush26agarwal)Photo by Rushil Agarwal (@Rush26agarwal)
technically from that perspective no one who has ever built a company should have the “right to evaluate” a company lol
tried to add some nuance to my take on why most of deeptech is a grift. hopefully this inspires more chatter around the ecosystem 🤪🤪🤪
1
355
Rushil Agarwal retweeted
Halfway through W27 interviews. Intense weeks for us and for the founders applying, and a privilege to hear how they plan to change the world. Many are still in school or just graduated. There is one school that consistently stands out this cycle: UC Berkeley.
96
56
54
1,321
111,275
Will be in Chile and Argentina end of November and beginning of December. Would love to meet more data collection partners in LATAM, DM me if you are scaling ego data collection. (Open to flying to more countries outside of Argentina and Chile)
1
2
10
1,225
We will be at CoRL this week, would love to talk to new researchers. Dm if you’re around!
1
7
888
I am hiring for a Research Study Coordinator (Operations, Part-time) at Human Archive (YC W26) San Francisco (onsite) · 20 hrs/week · $25-50/hour We’re running a research study in our SF office and need someone to manage the day-to-day operations behind it. It is one of the best opportunities to get hands-on experience at the frontier of physical AI/robotics. You will work directly with our research engs and cofounders and own a core part of this project. What you will do: - Source and schedule 1000s of unique data collectors - Keep track of diversity distribution and manage data quality - Write and enforce the SOPs that keep sessions consistent - Set up capture environments and track which scenes are done Who you are: - In school or graduated; either works (pref. Sophomore - recent grad) - Preferably someone good with people, extremely creative (non-technical is fine) - Available for 1–6 months. We’re flexible on the exact length. Apply Now: jobs.ashbyhq.com/humanarchiv…
6
3
1
32
3,295
Rushil Agarwal retweeted
office is on fire but @Ishan_Ahlu is always grinding 🚀🦾
1
3
14
723
We're launching an Entrepreneur in Residence (EIR) program at Human Archive (Global: Work out of any of our offices across US, South Asia, China or remotely,). We also have some reserved spots for founders from LatAm, MENA, Eastern Europe! the demand for real-world data is growing rapidly and we want to accelerate the infrastructure around it. We want to work with people who can build businesses to close gaps and solve problems that we’re unable to at the moment. - You build and own 100% of your business - We bring guaranteed demand for your products - We support you with legal, financial, and operational infrastructure I think this is one of the best opportunities to build a business today. You get to build around real customers and actual demand from day one in an industry that’s early yet growing It doesn't matter if you've built a company before. We care more about your ability to figure things out and execute. Apply now: docs.google.com/forms/d/e/1F…
16
11
2
267
21,982
Cant call a space cluttered when quality is scarce
We pre-trained the same VLA on 100 hours of Human Archive’s OTS dataset and 100 hours of randomly sampled data from each of @microagi’s and @MeckaAI’s datasets on EgoVerse. Overall, Human Archive data achieved the strongest results, with the highest overall task success rate. We observed particularly strong generalization to unseen objects. Thread below 🧵👇
3
20
2,454
Rushil Agarwal retweeted
We pre-trained the same VLA on 100 hours of Human Archive’s OTS dataset and 100 hours of randomly sampled data from each of @microagi’s and @MeckaAI’s datasets on EgoVerse. Overall, Human Archive data achieved the strongest results, with the highest overall task success rate. We observed particularly strong generalization to unseen objects. Thread below 🧵👇
24
33
6
170
16,110
Rushil Agarwal retweeted
We’re scaling compute for our perception models and commercial deployments. The team is actively looking for creative ways to partner. Please reach out!
3
2
17
1,419
Rushil Agarwal retweeted
One year ago from today, @Rush26agarwal @shloke_patel and I started thinking about human video/sensor data for robot learning!
3
1
42
2,796
Rushil Agarwal retweeted
Human data for robot learning is progressing quickly. Last November, there were <5 startups in this space. Now there are 100s of companies — some only building hardware, some not building devices or collecting data but doing “data quality,” some only collecting data using others’ hardware, and many claiming they do it all in-house. If you want your data to be good, you need to do it all in-house, benchmark your policies, and deploy robots :) We’re still in the early innings. Exciting releases coming later this month!
We benchmarked only 7 hours of Human Archive data against 4 widely used public datasets totaling 17 hours: HOT3D, H2O, HOI4D, and TACO. The results: Human Archive only: 85% task success | 65.6% task progress 4 public datasets: 57.5% task success | 52.5% task progress Mixture of all 5: 60% task success | 56.3% task progress
7
5
1
40
4,628
Rushil Agarwal retweeted
Never doubt an insanely young, technical, and ambitious team!
5
5
56
2,723
Rushil Agarwal retweeted
Pulled the fastest growing startups by hiring velocity over the past 90 days:
72
49
132
898
498,078
Rushil Agarwal retweeted
life mostly changes in seconds/moments, for better or worse the time between these moments is a result of the big moment preceding it and building up to the next big moment the pareto principle applies here as well: 80% of life is a result of 20% of the moments
2
3
234
I am hiring for five cracked partnerships/sales people in India We collect data across 20+ different categories, such as food and beverage, hospitality, industrial, construction, creatives, engineering etc. I need people who are: 1) Type A when it comes to communication, organisation, and planning 2) Understanding metrics and data well. Keeping track of diversity distribution for different categories and constantly hitting short-term and long-term targets is not possible without understanding analytics. So be goated at KPIs, analytics, etc. 3) Does not believe in slow sales cycles or "industry standards"; what we do has never been done in the history of humanity, so if you have any preconceived notions about how partnerships work, you need to get rid of them 4) Has a very particular niche such as enterprise/corporate sales or MSMEs or college networks or legacy businesses 5) Be shameless, have no ego, but know how to boost someone else's ego 6) Not hesitant about leveraging personal network We pay extremely well for data collection. It is money that is hard to turn down, so at the end of the day it comes down to if you can build distribution through channel partners and scale the hell out of converting leads. But most importantly, be fine with constantly hopping on calls, repeating the same pitch 100 different times. Partnerships team will be converting 100+ UNIQUE businesses a month across all categories. So speed, hunger and clarity are non-negotiable. 20-40 LPA + bonuses Liberty to build ur team (get interns, do whatever it takes) Ideally available to start immediately. INR 1 Lakh in referral bonus. jobs.ashbyhq.com/humanarchiv…
5
4
1
36
4,967