
🇵🇭 | recommendation systems @netflix | built @ploopyapp, @AcadiaAi | ai @CarnegieMellon | prev ads @meta, founding engineer/mle @fanhouse
Joined April 2020
- Tweets38
- Following490
- Followers165
- Likes153
nathan retweeted
as a design engineer I often get pinged for tiny front-end changes
at the @cursor_ai hackathon we used background agents to automatically convert design system updates into PRs 🎨✨
ft @_nathanmarquez_ @og_chamod @FateJacobson
Introducing ww.writeaway.fun 📝
An AI native document editor - like Cursor + Notion!
- Tab to complete
- Use the AI Agent to suggest and make edits to docs
- "AI Blocks" that update content reactively
Super fun doing this for @boltdotnew's world's largest hackathon :)))
nathan retweeted
squishy ploopy is here!!🥹
finally brought these little guys to life with animation 🏃 and a redesigned ploopy profile ✨
lmk if you’re interested in getting your own animated ploopy!
#animation #digitalart cc: @_buildspace @FarzaTV @ploopyapp
joined @_buildspace with @lemonyummyart! we're building PLOOPY 🤩☁️💡
give your brain and relationships +10000 aura
will be posting updates here as we build out this silly little app - follow to stay updated :)
cc: @_buildspace @_nightsweekends
nathan retweeted
Here's a new SOTA text-to-image eval metric that's much better at complex compositional reasoning than current ones (e.g CLIPScore, PickScore)!
We also show that it generalizes to video/3d evaluation + released a comprehensive t2visual meta-eval metrics benchmark.
Great to have worked on this with w/ @ZhiqiuLin, @deepakpathak, @gneubig, @PengchuanZ, @RamananDeva. Paper + code ⤵️
In text-to-image generation, evaluating how well the generated image matches the prompt is a major challenge. We address this with VQAScore: a SOTA metric that significantly surpasses CLIPScore, PickScore, ImageReward, TIFA, and more!
VQAScore works especially well on complex text prompts that involve attributes, relations, and reasoning. VQAScore also generalizes to video and 3D evaluation!
Code: github.com/linzhiqiu/t2v_met…
Paper: huggingface.co/papers/2404.0…
Work led by CMU & Meta with @ZhiqiuLin, @deepakpathak, @EmilyLiJiayao, @PengchuanZ, @gneubig, @RamananDeva
nathan retweeted
Super excited to introduce 🌳Acadia (@AcadiaAI ) Playground, an interpretable data exploration tool to understand your evaluation data’s quality and help unlock insights into model performance using AI!
🧵
nathan retweeted
Problems I'd do if I'm to do a startup again (though I probably won't any time soon because startups are hard). If you’re solving any of them, I’d love to chat.
1. Data synthesis: AI has become really good both at generating and annotating data. The challenge now is to make sure that the generated data is safe and legal, e.g. not violating any IP.
2. Evaluation: evaluation has gotten so much harder with LLMs, both because many people treat models as blackboxes (we deploy models someone else developed for us) and because outputs can be open-ended. At the same time, investment in evaluation is nowhere close to investment in model or application development. I’d be interested in arena-style evaluation, embedding evaluation, human-in-the-loop evaluation, as well as small, specialized scorers (instead of using large models like GPT-4 as judges).
3. Energy: the bottleneck to scaling AI is no longer compute but electricity. I’m interested in all energy-related problems, including both new energy sources and energy trading.
4. Any application that allows you to collect unique data that nobody has. I’ve heard concerns about building applications that seem to be “wrappers” around popular APIs. If you can get to the market early and gather sufficient data to continually improve your product, data is your moat.
5. GPU-native everything: many data science toolings, including scikit-learn, pandas, and Spark, aren’t built to run natively on GPUs. There have been efforts to make these tools more efficiently leverage GPUs, but I think there’s still a lot of room for the software layer for GPUs (and not just NVIDIA GPUs).
6. Curated Internet: bots are already ruining dating apps, search (bots are incredibly good at SEOs), and social media. I’d like to be able to set a boundary for my Internet, e.g. to limit the search results to those written by people I trust, or sources verified to be human.
nathan retweeted
3/ Hierarchical semantic clustering
Clustering scheme that generates an interconnected hierarchy that links ideas together into a single post
Consolidate your notes into a blog post
🥇First place
@JvNixon @_nathanmarquez_ @zvhgpyxqtnys
introducing BODEGA
an aggregated shopping feed app where everything is on sale
me and my brother nico built this for @_buildspace nights & weekends. check out or demo vid below + sign up for the beta at shopbodega.app/beta :)
cc: @_nightsweekends
youtu.be/RN56ITZxzy8?feature…
4
298
nathan retweeted
Check out this video our intern, Nathan, made about working at Fanhouse 🥳 Nathan is a rising sophomore at Carnegie Mellon, and at Fanhouse he's been working hard to help us build features (like preventing leaks)! 😊 Best intern we've ever had 💗
youtu.be/rNrYNV1C6mw
















