Alejandro (Alex) Chacon (@achacond) header
Alejandro (Alex) Chacon (@achacond) profile picture

AI Devtech at @NVIDIA. Former @XilinxInc, @Arm & @UABBarcelona. (AI, Genomics, HPC, GPUs/FPGAs/Embedded, Computer Architecture) Opinions are my own.

Cambridge, England
Joined May 2013
Photo by Alejandro (Alex) Chacon (@achacond)Photo by Alejandro (Alex) Chacon (@achacond)Photo by Alejandro (Alex) Chacon (@achacond)Photo by Alejandro (Alex) Chacon (@achacond)Photo by Alejandro (Alex) Chacon (@achacond)Photo by Alejandro (Alex) Chacon (@achacond)Photo by Alejandro (Alex) Chacon (@achacond)Photo by Alejandro (Alex) Chacon (@achacond)Photo by Alejandro (Alex) Chacon (@achacond)Photo by Alejandro (Alex) Chacon (@achacond)
Ending PhD @enginyeriauab, my last 7 years work in a plot, strong algorithmic & optimizations for a CPU-GPU read mapper. Opensource, free & used in production @cnag_eu GEM3-GPU goo.gl/pB5RTf GPU corelib goo.gl/znCdMM Thanks @nvidia & friends for the support!
4
8
1
43
Alejandro (Alex) Chacon retweeted
Chip design has been an art that took practitioners years to master. As with other areas of deep expertise, we're seeing the benefits of ML in this space and we have rapidly moved from proof-of-concept to actual deployment on hardware that matters! So excited to partner with @AMD to demonstrate the power of our models for better, faster, cheaper AI infrastructure. @OpenAI openai.com/index/advancing-c…
10
41
7
413
21,001
Alejandro (Alex) Chacon retweeted
I've had months to prepare for my University of Cambridge talk. When am I actually preparing it? On the train down of course! talks.cam.ac.uk/talk/index/2…
7
3
69
2,619
Alejandro (Alex) Chacon retweeted
🆕 cuPotential A CUDA-X library for computing nonbonded interactions in molecular dynamics. The OpenMM plugin replaces the nonbonded force, integrator, and barostat. In NPT benchmarks on systems from 24K to 400K atoms, it ran over 3x faster than OpenMM on RTX PRO 6000 Blackwell and about 2x faster on B200, H100, A100, L40S, A10, L4, and T4. Accuracy holds up too, with a few hundred times less energy drift than default PME. 💻 Github nvda.ws/3WIwogJ 📄 Technical Blog nvda.ws/4AyIpDX 📍 Find us at #SC26 👇
5
78
19
428
35,395
Alejandro (Alex) Chacon retweeted
Gemini for Healthcare!
Today, we’re announcing Gemini for healthcare. This newest addition to our suite of purpose-built industry solutions extends the Gemini agent to health systems and health plans. Learn more → goo.gle/4zf0s11
6
9
99
19,379
Alejandro (Alex) Chacon retweeted
🧪 @UWproteindesign + 💚 NVIDIA = 👶 RFD4-Proteina Proud parents ✨ Gifted kid
4
53
10
237
24,127
Alejandro (Alex) Chacon retweeted
Proud to share an important milestone: the first prospective study evaluating AMIE, our research conversational medical intelligence, with real patients in an active clinic is published in @TheLancet The Lancet flagship journal is one of the world’s oldest and highest-impact medical journals. It marks a meaningful step in our research journey, building on foundations with Med-PaLM and milestones with AMIE. Key results in the next post 🧵👇
4
15
1
90
6,298
Alejandro (Alex) Chacon retweeted
1. Excited to see what CZI & teams will do. These are great places with A+ people, data & compute infra w/ real resources (~5X ENCODE in ~5x less time). Problem is imp. enough to spend that much (~a workweek of iphone sales) 2. My wariness remains: srikosuri.substack.com/p/the…
BREAKING: The U.S. government, Google, Meta & Zuckerberg-backed @biohub are launching a $1.8B push to build the data layer for AI biology. The goal is to create the largest open, AI-ready biological dataset ever assembled, giving models the data needed to better predict how biology works & ultimately help treat disease. The coalition is massive: › Biohub committed $500M › DOE is investing $500M+ across measurement, modeling & compute › NIH is bringing $500M+ of federally funded datasets & repositories › Meta, Google DeepMind & @IsomorphicLabs are jointly investing another $300M This is effectively an infrastructure buildout for AI biology. Instead of just building bigger models, they’re attacking one of the biggest bottlenecks in biology AI today, generating standardized, high-quality biological data at massive scale & making it openly available to researchers. $1.8B committed to turning biology into something AI can increasingly model, predict & engineer.
1
7
1
66
8,929
Alejandro (Alex) Chacon retweeted
🧬 From protein structure prediction to designing high-affinity binders Explore how NVIDIA’s BioNeMo Agent Toolkit brings AI-driven drug discovery into practice. Free on-demand ▶️ nvda.ws/3TE1Ifr
3
19
3
102
4,055
Alejandro (Alex) Chacon retweeted
The most important thing we can apply AI to is improving medicine and human health - very excited about this partnership with the CZI and the Virtual Biology Initiative!
Today, Isomorphic Labs joins the Virtual Biology Initiative (VBI) as a founding member. We are helping build an open resource for the global research community to accelerate how we understand and treat disease. Read more here: isomorphiclabs.com/articles/…
238
516
58
5,780
241,464
Alejandro (Alex) Chacon retweeted
This weekend was fun, as I now have a PacBio DeepVariant caller with < 1 minute TAT 🎉 Even pure CPU is faster than what others are offering on GPU.
This weekend's Opus /goal was simple and crazy. Run DeepVariant variant calling on a 128c + 1 RTX 6000 Blackwell server in less than 60s.
4
2
1
17
1,719
Alejandro (Alex) Chacon retweeted
Introducing AlphaProtein Novo, a pipeline for de novo enzyme design! In collaboration with @francesarnold's group, we developed AP Novo and used it to get SoTA "novel-scaffold" activity on 2 benchmark reactions and create enzymes to synthesize the pharmaceutical motif piperidine and degrade the environmental toxin DEHP. Work led by @ZvxyWu @ZiqiLi0513 with teammates @alexechu_ @kelly_ruijie @schulluc @jacobjinkelly @rosaliags @davidla @solserpens @j_reisenbauer, @pengliu_group, Josh, Thomas, Amy, Tristan, Harshnira, and others. It was a joy and privilege to work with you all! Preprints and code: biorxiv.org/content/10.64898… biorxiv.org/content/10.64898… github.com/google-deepmind/a…
20
179
24
1,024
139,725
Alejandro (Alex) Chacon retweeted
Google DeepMind just moved protein AI from reading structures to designing chemistry. Its new system, AlphaProtein Novo, generates de novo enzymes for catalytic reactions — including enzymes that can synthesize drug-like piperidines and degrade the plasticizer DEHP, even at 90°C. The bigger story: AlphaFold learned to read proteins. AlphaProteo learned to design binders. Now AlphaProtein Novo is beginning to design catalysts. From predicting biology → writing proteins → engineering chemistry. Preprints and code: biorxiv.org/content/10.64898… biorxiv.org/content/10.64898… github.com/google-deepmind/a…
Introducing AlphaProtein Novo, a pipeline for de novo enzyme design! In collaboration with @francesarnold's group, we developed AP Novo and used it to get SoTA "novel-scaffold" activity on 2 benchmark reactions and create enzymes to synthesize the pharmaceutical motif piperidine and degrade the environmental toxin DEHP. Work led by @ZvxyWu @ZiqiLi0513 with teammates @alexechu_ @kelly_ruijie @schulluc @jacobjinkelly @rosaliags @davidla @solserpens @j_reisenbauer, @pengliu_group, Josh, Thomas, Amy, Tristan, Harshnira, and others. It was a joy and privilege to work with you all! Preprints and code: biorxiv.org/content/10.64898… biorxiv.org/content/10.64898… github.com/google-deepmind/a…
7
83
8
564
29,879
Alejandro (Alex) Chacon retweeted
Five pharma companies just fine-tuned OpenFold3 in a federated setting on 20,000+ structures... ... raising correct ligand-pose predictions from 29% to 47%
5
44
5
267
25,721
Alejandro (Alex) Chacon retweeted
Today, we're sharing some of the "secret sauce" behind the acceleration in NIM microservices, as an open-source library for anyone to use It's saved us a ton of compute / time / $ when generating protein structures at million-scale for the AlphaFold Database So check out BioNeMo Inference Runtime as a public beta release today! 🧵
1
9
66
3,520
Alejandro (Alex) Chacon retweeted
I'm so excited about this BioNeMo inference runtime for faster protein structure prediction! NVIDIA is making such an incredible impact in the BioAI field & making it all open source, it's as good as it gets! This will actually save lives, unlike all the doomer virtue signaling!
👋 Say hello to faster structure prediction NVIDIA BioNeMo Inference Runtime is in public beta. This open, PyTorch-native library speeds up biomolecular inference on NVIDIA GPUs. Optimize models including Boltz-2, OpenFold2, and Protenix v2 with specialized kernels and, where supported, CUDA Graphs - then scale with Ray-powered GPU replicas. 🙌 Huge thanks to our collaborators at @apheris_AI, @AurekaBio, @Biomap_IMI, @boltz_bio, @dataiku, @UWproteindesign, @Latent_Labs, MoleculeMind, @nebiusai, @openmsf, @open_fold, @proximabio, @SandboxAQ, @Terray_Tx, and @Xaira_Thera for partnering with us on this work. 📃 nvda.ws/46PDmlL
2
27
1
292
22,593
Alejandro (Alex) Chacon retweeted
One of our most important releases of all time. We’ve accelerated a lot of models and written tons of advanced kernels — but figuring out how to generally and flexibly accelerate the wide variety of structure-based models used for inference has been a huge challenge. Benchmarked, documented, open sourced. Get in touch and tell us how it’s going and how we can improve. Congratulations to our engineering, research, devtech teams, @TadimetiNeha and (not on twitter) Mahan Salehi in product management! Just the beginning!
👋 Say hello to faster structure prediction NVIDIA BioNeMo Inference Runtime is in public beta. This open, PyTorch-native library speeds up biomolecular inference on NVIDIA GPUs. Optimize models including Boltz-2, OpenFold2, and Protenix v2 with specialized kernels and, where supported, CUDA Graphs - then scale with Ray-powered GPU replicas. 🙌 Huge thanks to our collaborators at @apheris_AI, @AurekaBio, @Biomap_IMI, @boltz_bio, @dataiku, @UWproteindesign, @Latent_Labs, MoleculeMind, @nebiusai, @openmsf, @open_fold, @proximabio, @SandboxAQ, @Terray_Tx, and @Xaira_Thera for partnering with us on this work. 📃 nvda.ws/46PDmlL
6
14
112
7,578
Alejandro (Alex) Chacon retweeted
👋 Say hello to faster structure prediction NVIDIA BioNeMo Inference Runtime is in public beta. This open, PyTorch-native library speeds up biomolecular inference on NVIDIA GPUs. Optimize models including Boltz-2, OpenFold2, and Protenix v2 with specialized kernels and, where supported, CUDA Graphs - then scale with Ray-powered GPU replicas. 🙌 Huge thanks to our collaborators at @apheris_AI, @AurekaBio, @Biomap_IMI, @boltz_bio, @dataiku, @UWproteindesign, @Latent_Labs, MoleculeMind, @nebiusai, @openmsf, @open_fold, @proximabio, @SandboxAQ, @Terray_Tx, and @Xaira_Thera for partnering with us on this work. 📃 nvda.ws/46PDmlL
23
210
26
1,133
204,772
Alejandro (Alex) Chacon retweeted
🧬 From sequence to structure - no manual glue code required. We've integrated NVIDIA BioNeMo Agent Toolkit into @AnthropicAI Claude Science, letting agents orchestrate MSA generation + multi-model structure prediction (OpenFold3, Boltz-2) end-to-end from a natural language prompt. 🛠️ The BioNeMo Agent Toolkit gives agents domain-specific life sciences skills, and on internal benchmarks it lifts task correctness from 60% to 100% while ~doubling token efficiency 🔬 In a real test case (Seh1 + a predicted fungal partner protein), the agent ran MSA search, then predicted structures with two independent folding models, preserving every request, response, and confidence score for review 📊 MSA is the load-bearing input: heteromer interface confidence (iPTM) hit 0.85 (OpenFold3) and 0.82 (Boltz-2) with alignment data, but collapsed to 0.14 and 0.19 without it Details nvda.ws/4gtXFLf
14
45
5
230
22,664
Alejandro (Alex) Chacon retweeted
A PhD student in the Baker Lab is using Claude to take over the 4 a.m. lab shift. Zihao Song connected six instruments in under a week using Anthropic’s Model Hardware Standard (MHS), letting an AI agent monitor qPCR in real time, stop reactions at the right moment, coordinate robotic plate handoffs, and oversee experiments remotely. AI can already design proteins. Now it’s starting to operate the wet lab too. anthropic.com/news/model-har…
🤖 Made with AI
11
44
4
422
74,816
Alejandro (Alex) Chacon retweeted
Using NVIDIA Proteina-Complexa, @muni_bio's autoresearch agent explored nearly 14,000 protein designs. @adaptyvbio validated nine TREM2 binders, three with sub-nanomolar affinity. The results show how agents can connect computational design with wet-lab feedback to improve the next round of discovery. 📘muni.bio/research/closing-th…
2
25
7
103
15,644