Ava Amini
@avapamini
Followers
9K
Following
2K
Media
24
Statuses
382
principal researcher @MSFTResearch | AI for biomedicine | instructor @MITDeepLearning | alumna @MIT @Harvard
Joined September 2014
thrilled to share The Dayhoff Atlas of protein language data and models π protein biology in the age of AI! https://t.co/4wP9kNRUoM we built + open source the largest natural protein dataset, w/ 3.3 billion seqs & a first-in-class dataset of structure-based synthetic proteins
4
78
311
Interested in developing generative models leveraging a diversity of architectures, modalities, and techniques? MSR is hiring Machine Learning Engineers in Cambridge, MA. Come join some of the most brilliant, kind, humble individuals Iβve ever encountered!
1
6
11
This Spring, Your Future Awaits! Enter today for a chance to win your share of $250,000 in scholarships. Be one of the 50 winners to unlock new possibilities!
0
16
146
Last week we (@LiquidAI) announced Liquid Labs! π§ What is Liquid Labs? Here's your answer: Read more: https://t.co/ez46Zh1b62 Video credit: @mlech26l @ramin_m_h @loo_noel @mihirbafna14 @961014dltkdg
1
7
45
π’ Registration of the NINTH (!) year of @MITDeepLearning has officially opened β with over 15M registered students taking the course over the past 9 years. And this this is just the beginning! Sign up TODAY to join the 2026 edition π https://t.co/apU376yAG9
0
7
12
Today we introduce Liquid Labs, our advanced research unit, with the goal of understanding and building efficient and adaptive intelligence systems. Liquid Labs consolidates our existing research efforts at Liquid across architecture of foundation models, multimodality,
18
34
241
The @LiquidAI LFM2 Tech Report is now live! It's a 51-page behemoth π with the full recipe on how we pre-, mid-, and post-trained LFM2, across all modalities text, vision, and audio π π https://t.co/Fm90fRdLTa π€
huggingface.co
The LFM2 Tech Report is now live on arXiv! We share everything from our novel hardware-in-the-loop architecture design, pre-training, and knowledge distillation, to the post-training recipe for small models. > π€LFM2 class of models has over 3.3M downloads > βοΈLFM2 nanos from
2
6
23
Excited to share the preprint from my @MSRNE internship project: Adaptive resampling for improved machine learning in imbalanced single-cell datasets Preprint: https://t.co/AlmFiTj7hl GitHub:
github.com
Adaptive resampling for machine learning in single-cell biology - microsoft/sc-AR
single-cell models tend to learn from the many - and miss the rare we introduce an Adaptive Resampling approach to help models learn from underrepresented cells, improving generalization & discovery https://t.co/CIpAW1RwWA
https://t.co/xfCc13dMe5 great work by @NavidiZeinab!
1
3
7
single-cell models tend to learn from the many - and miss the rare we introduce an Adaptive Resampling approach to help models learn from underrepresented cells, improving generalization & discovery https://t.co/CIpAW1RwWA
https://t.co/xfCc13dMe5 great work by @NavidiZeinab!
1
16
100
Bring soft artistic beauty to your space with this 135-piece cat-shaped Morandi puzzle.
0
1
2
We have recently uploaded our revised manuscript βEvaluating the role of pre-training dataset size and diversity on single-cell foundation model performanceβ. TL;DR: More models, more tasks => same results.
1
3
4
Boston is so different man. At Area Four and there are like 3 people doing deep mutational scans reliant on high throughput readout pipelines. You gotta be here
SF is so different man. At the local coffee shop and there are like 3 people coding. You gotta be here
8
13
328
Are you a PhD student interested in machine learning and biology or health? Come do an internship with me, @avapamini, @alexijielu, @lorin_crawford, or Kristen Severson at @msrne! Applications are due Dec 1: make sure you include a research statement! https://t.co/HfyZshDtof
12
106
458
Don't miss this great opportunity to work with the MSR BioML team. I was mentored by @avapamini and @KevinKaichuang and highly recommend it. The undergrad research internship deadline is tomorrow (Oct 6), so apply right now!
Applications for @MSFTResearch undergrad research internships for rising juniors and seniors are due Monday Oct 6! Apply to work with us in BioML π https://t.co/QYjsdyEpRQ w/ @KevinKaichuang, @alexijielu, @lorin_crawford, Kristen Severson, @ntenenz, @SarahAlamdari, and more!
0
5
36
Tired of losing things between your car seats? πCrumbs, keys, and toys disappear? It's over now! Meet the Car Seat Side Gap Filler β your tidy solution! Stay organized and drive stress-free. Grab yours now!
10
47
439
Applications for @MSFTResearch undergrad research internships for rising juniors and seniors are due Monday Oct 6! Apply to work with us in BioML π https://t.co/QYjsdyEpRQ w/ @KevinKaichuang, @alexijielu, @lorin_crawford, Kristen Severson, @ntenenz, @SarahAlamdari, and more!
3
23
125
Introducing @Liquid_AIβs first "omni-modal" model β with the ability to both hear and talk! ππ β
One end-to-end foundation model. β
Seamless support across many tasks. β
All on-device, no API costs! Read more: https://t.co/2lSgQ5Odmb Huggingface:
huggingface.co
Today, we expand our LFM2 family to audio. ππ LFM2-Audio is an end-to-end audio-text omni foundation model, and delivers responsive, real-time conversation on-device at just 1.5B parameters. One model. Seamless multimodal support. No chains. > Speech-to-speech >
1
5
24
Today, we expand our LFM2 family to audio. ππ LFM2-Audio is an end-to-end audio-text omni foundation model, and delivers responsive, real-time conversation on-device at just 1.5B parameters. One model. Seamless multimodal support. No chains. > Speech-to-speech >
24
77
468
When it comes to deep learning for protein engineering, there is strength in simplicity. In a Preview piece @CellCellPress, we highlight work from @CaixiaGaoLab on using fixed-backbone sequence design to engineer genome editors with improved function. π https://t.co/NTjDMY4RP6
1
26
148
Exciting work by our team, representing a real step forward in understanding the protein dynamics that power biological function.
Researchers have developed a #DeepLearning system called BioEmu that rapidly generates diverse protein conformations, enabling fast, accurate insights into protein flexibility and function. Learn more this week in Science: https://t.co/Pe15hm9F52
75
205
2K
AMD π€ @LiquidAI_ = Faster, smarter, on-device AI.
Big step for on-device AI: Liquid AIβs Edge Platform, LEAP, now supports @AMD Ryzenβ’ and Ryzen AIβ’ processors, bringing powerful, low-latency AI directly to laptops. Hereβs what it means for developers and enterprises π§΅
6
43
222
Excited to share our work in @CellCellPress on using generative AI to design entirely new antibacterial molecules that kill Neisseria gonorrhoeae and Staph aureus! https://t.co/wnJkDWEVtn
@MGHPath @mgh_id @broadinstitute @MIT #AMR
3
7
19
Today @LiquidAI_ released LFM2-VL! πΈ - super fast VLMs in two sizes 450M and 1.6B π₯ - optimized for all kinds of edge devices π±/π/π€/ π»... - native resolutions and aspect-ratios π - open-weight and downloadable TODAY π€ https://t.co/17jvR0JM4N
7
43
228
Modern LM training is a game of π± and π. You improve training signal (e.g., data) to overcome gaps in eval performance, only to strengthen the evals and thus discover new gaps -- starting the cycle anew. With the Dayhoff Atlas, we aim to jumpstart the same race for PLMs. We
1
7
25
Experience comfortable reading with lenses designed to reduce blue-light exposure and enhance visual clarity. The dual-focus design supports both near and mid-distance viewing, helping you read and work with ease throughout the day.
0
24
142