012024–26
Robots that learn from a single demonstration
A compact dexterous policy that can adapt after one human example.
RoboticsImitation learning
A personal research page, without rebuilding a website
Research deserves more than a citation list.
Turn papers, projects, methods, figures, and ideas into one research page that makes the work easier to understand — and easier to remember.
Free to startNo card requiredChoose the final address before publishing
Published profile previewrestack.my/mayapatel
About
My group studies how robots can acquire useful behavior from a handful of human demonstrations. We combine tactile sensing and human-centered evaluation.
Featured research
Publications
RSS · 2026
FeaturedOne-shot policy learning through tactile priors
M. Patel, J. Kim, A. Rodriguez
A tactile prior reduces the demonstration data needed for dexterous policy learning.
RoboticsTactile learning
Research in circulation
A page worth publishing should be easy to share.
This space is reserved for approved X posts from researchers. For now, the feed below uses fictional ReStack profiles to preview how real snapshots will sit in the page.
Maya Patel@mayapatel_lab·2h
DemoPut our tactile-learning work into one page people can actually send around — paper, figures, project context, all together.
184128612K
Jonah Lee@jonahcoast·4h
DemoNew coastal adaptation field atlas is live. The project page finally feels easier to update than the site around it.
7221436.8K
Ari Nwosu@arinwosu·1d
DemoA portfolio where the project story and the evidence live next to each other. This is closer to how I actually want people to encounter the work.
11352019.4K
Demo snapshots only. No fictional post above is presented as a real endorsement; production social proof remains limited to approved posts with preserved source links.
Research, without the web maintenance tax
Your research is more than its title.
A personal site can be powerful. ReStack just moves the center of gravity: update the research once, and let the page, address, and search layer move with it.
Maintain the website
one research update touches several web concerns
new research
Tactile priors for one-shot manipulation
GitHub Pages
Vercel
Netlify
NotionMaintain the research
the web layer follows
eStack
Maya Patel
restack.my/mayapatel
One-shot policy learning through tactile priors
search-ready
restack.my/mayapatel
Maya Patel — Robotics Research
metadata follows the research
add
publish
discoverable
The invisible web work, handled
Publish the research. ReStack handles the web layer.
Technical SEO is a stack of small, easy-to-miss settings. ReStack gives every published profile a clean address and prepares the search, structured-data, sitemap, and social metadata around it by default.
Usually hidden in web config
Search readiness is more than one SEO field.
config
Title + description
config
Canonical URL
config
sitemap.xml
config
robots.txt
config
Person JSON-LD
config
Open Graph + social
ReStack maintains this layer.
No config files to keep in sync for a normal published profile.
One publish
Update the research, not the SEO stack.
Search preview
generatedrestack.my/mayapatel
Maya Patel — Robotics Research
Tactile learning, one-shot policy learning, and human-guided manipulation.
Human-friendly address
restack.my/mayapatel
Stable, readable, and used as the canonical profile URL.
Social metadata
A useful preview when the page is shared.
Open Graph and large-card metadata travel with the profile.
Canonical profile URL
Published profile in sitemap
Person structured data
Search + social metadata
Automatic defaults on published profiles.Pro · customize SEO title + description
ReStack prepares the technical search layer; search ranking still depends on the quality, relevance, and reach of the research itself.
Research people can follow
Tell the story. Keep the evidence close.
A citation proves the work exists. ReStack keeps the question, method, result, figure, links, and publication context together so someone can actually follow the research.
Fig. 02 · transfer across unseen objectsIllustrative research demo
- Question
- Can a robot learn a useful manipulation policy from one human demonstration?
- Method
- Combine tactile priors with a compact imitation-learning policy and test on unseen objects.
- Result
- The policy transfers with fewer examples while keeping the experiment legible to non-specialists.
Smart Import
Existing Studio flow
DOIDOCX
Find a publication
10.1234/example.2026.184
ImportPublic metadata found · manuscript enrichment available when the source permits
Public metadataAI drafted
Tactile priors for one-shot manipulation
Question
Approach
Contribution
Review every field before savingOpen in Studio
Add research without rebuilding
Start from the paper you already have.
Smart Import accepts a DOI or DOCX manuscript, drafts the structured research fields, brings in figures when available, and keeps every field editable before saving.
DOI lookup with public metadata
DOCX manuscript import up to the current product limit
Editable Question · Approach · Contribution draft
See what people care about
A research page can tell you something back.
Private analytics show profile views, unique visitor-days, and link clicks. Pro adds aggregate referrer and country context without turning the page into an ad tracker.
Demo values only · analytics remain private and cookie-less.
Last 30 days
Private analytics
184
Profile views
127
Unique visitor-days
38
Link clicks
Top referrers · Pro
scholar.google.com42%
direct31%
university.example17%
What this tells you
See whether people arrive from search, shared links, or institutional pages — and whether they continue into your work.
Your research. Your visual language.
Same work. Different voice.
Switch the page around the research, not the research around a template. These previews use ReStack's real palette, typography, width, sidebar, surface, and card semantics.
Warm serif hierarchy with a balanced rail and figure-first research cards.
Sage paletteClassic typeFull bleedFigure-first
About
My group studies how robots can acquire useful behavior from a handful of human demonstrations. We combine tactile sensing and human-centered evaluation.
Featured research
012024–26
Robots that learn from a single demonstration
A compact dexterous policy that can adapt after one human example.
RoboticsImitation learning
Publications
RSS · 2026
FeaturedOne-shot policy learning through tactile priors
M. Patel, J. Kim, A. Rodriguez
A tactile prior reduces the demonstration data needed for dexterous policy learning.
RoboticsTactile learning
From blank page to shared page
Three steps. No site rebuild.
ReStack is designed to stay out of the way: claim the page, add the work, publish the version you want people to see.
- 01
Claim your page
Choose a memorable restack.my address and create your researcher identity.
restack.my/mayapatel - 02
Add your work
Import a DOI or DOCX, or add projects and publications manually in Studio.
DOI · DOCX - 03
Publish and share
Preview the page, publish when it is ready, and share one URL wherever your work travels.
Published
restack.my/mayapatel
Simple by default
Start free. Grow when the page earns it.
Publish a thoughtful research page without paying first. Pro removes research limits and opens deeper design, analytics, and SEO controls when you need them.
Free
$0forever
- Up to 3 projects
- Up to 10 publications
- Core design controls
- Basic private analytics
Pro
$8/ month
or $72 billed yearly
- Unlimited research
- Premium design controls
- Views, clicks & referrers
- Custom SEO · no ReStack branding
Your page is one address away
Put the work somewhere people can remember it.
Claim the address now. Add the research when you are ready.