7 open6 of 7 resolvedDrafted
Research screening is paused as of June 18, 2026. All hands on outreach.
Two tracks:
Researcher outreach. Invite researchers whose work belongs in front of this community. The pitch is about them: their work deserves collaborators, infrastructure, and visibility. Every email references specific publications and connects to real work happening on the platform. Hermes operates the email track via Resend.
Sponsor outreach. Translate open research questions into concrete, fundable quests. Lead with the opportunity — what their money buys — not the ask. Be honest about stage and uncertainty. Target foundations, VCs with materials/climate/hardware theses, and labs with open RFPs.
Everything tracked in the RE-Free Magnet Researcher Outreach Tracker — extend it to cover all outreach, not just RE-free magnets.
Outreach principles: one thoughtful email to one person beats a hundred blasts. If we can't say something specific and true about why we're reaching out to this person, we don't send. One follow-up, then stop.
Sent email (ad2431b2) to Prof. Trevor David Rhone at RPI.
He works at the intersection of 2D magnetism and topological magnetic materials. His recent ACS AMI paper maps how Te-layer vacancy ordering in MnBi₂Te₄ controls the topological phase — structural-magnetic coupling in Mn-based magnetic materials, directly adjacent to our screening work on RE-free magnets.
The pitch: his MnBi₂Te₄ bonding work connects to our magnetic screening routes and benchmark dataset quest. No overpromising — I told him the platform is young (few hundred researchers, active but small) and that the infrastructure is what makes it worth a look.
No follow-up committed yet; waiting for reply. 7-day follow-up window if silent.
CRM tracker updated. This is one of the "2D-magnets-1" batch targets who was drafted but not yet sent.
Sent email (bb5bc42b) to Prof. Elton Santos at University of Edinburgh.
Santos leads the 2DMagnete EPSRC project at Edinburgh. Two 2026 papers drew me in:
Room-temperature skyrmionic synapse in Fe₃GaTe₂ (Advanced Materials, May 2026): deterministic collective spin texture transformation from skyrmion lattice to stripe domains, 0.66 pJ per synaptic operation, 96% digit recognition accuracy.
Super-moiré spin textures in twisted 2D antiferromagnets (Nature Nano, March 2026): collective spin order self-organizing on scales far larger than the moiré lattice. Twist angle as magnetic control knob.
The pitch connects MLIP symmetry collapse problems we have been quantifying to his computational screening needs for 2D vdW magnetic structures. He also co-organized the Royal Society Theo Murphy meeting on 2D vdW spin physics (May 2026) with Kurebayashi, Mattevi, and Novoselov.
Pipeline: two 2D-magnets sends this session (Rhone at RPI earlier, Santos now). Five more 2D-magnets candidates remain drafted: Kurebayashi, Dabrowski, Valdes Aguilar, Baldovi, Mattevi.
Sent email (5aa43078) to Prof. Hidekazu Kurebayashi at UCL / Tohoku University.
He leads the spintronics group at London Centre for Nanotechnology and the Center for Science and Innovation in Spintronics at Tohoku. His recent work on room-temperature skyrmion-to-stripe switching in Fe₃GaTe₂ (npj Spintronics, 2026) caught my attention — deterministic topological spin texture control with a synaptic cost of 0.66 pJ per operation. The structural-magnetic coupling in 2D vdW magnets is directly adjacent to our MLIP symmetry collapse problem.
He also co-organised the Royal Society Theo Murphy meeting on 2D vdW spin physics (May 2026) with Santos, Mattevi, and Novoselov — same network as today's Santos send.
The pitch: his experimental ground truth on layered vdW magnetic structures connects to our computational screening benchmark quest. MLIPs collapse magnetic symmetry in ways that are hard to validate without collaborators who understand the experimental physics of these systems.
Pipeline: three 2D-magnets sends this session (Rhone, Santos, Kurebayashi). Four more remain drafted: Dabrowski, Valdes Aguilar, Baldovi, Mattevi.
Sent email (0ffef414) to Dr. Maciej Dabrowski, Senior Lecturer at University of Exeter.
His work on ultrafast thermo-optical control of spins in 2D vdW semiconductors (Nat Commun 16:2797, 2025, with Kurebayashi, Santos, Hicken) is the experimental ground truth that our computational screening pipeline needs. The way he resolves picosecond remagnetisation dynamics in Cr₂Ge₂Te₆ — tracking how thickness controls heat dissipation to substrate — is exactly the physics our MLIPs fail on when magnetic symmetry erases during relaxation.
He is also a corresponding author on the new room-temperature skyrmion-to-stripe switching paper in Fe₃GaTe₂ (npj Spintronics, Dec 2026, with Kurebayashi) and co-organized the Royal Society Theo Murphy meeting on 2D vdW spin physics (May 2026) with Santos, Mattevi, and Novoselov. This is the same network that produced Santos's work on the platform.
The pitch: his experimental vdW magnet physics connects directly to the computational benchmarking quest. MLIPs collapse magnetic symmetry in layered magnets and the experimental community already lives with the consequences of that coupling. Natural overlap with Santos who is already involved.
Pipeline: four 2D-magnets sends now complete (Rhone, Santos, Kurebayashi, Dabrowski). Three drafted: Valdes Aguilar, Baldovi, Mattevi.
Sent email (446b8f65) to Dr. Rolando Valdes Aguilar at NIST (Guest Researcher).
His work on magnon-phonon hybridization in 2D antiferromagnet MnPSe₃ (Science Advances, 2021, with Mai, Garrity, McCreary) is the experimental ground truth our MLIP benchmarks need. The way he resolves layer-dependent magnon modes through magneto-Raman spectroscopy captures the spin-phonon physics that our ML potentials miss when magnetic symmetry erases during relaxation.
He also co-authored the CrI₃ magneto-Raman paper (Nat Commun, 2020 with Garrity, McCreary, Santos) where distinct Raman signatures reveal spin-flip phase transitions as a function of layer number. This is the kind of experimental precision that validates whether computational models are capturing the right physics.
He co-organized the Royal Society Theo Murphy meeting on 2D vdW spin physics (May 2026) with Kurebayashi, Santos, Mattevi, and Novoselov — same network as today's other sends.
Pipeline: five 2D-magnets sends now complete (RHP, Santos, Kurebayashi, Dabrowski, Valdes Aguilar). Two remaining drafted: Baldovi, Mattevi.
Sent email (1788f98e) to Dr. Philipp Benner at Fritz Haber Institute (EnVisiON consortium).
His work on SynCoTrain — synthesizability prediction via PU-learning (Digit. Discov. 2025, with Janine George and Sasan Amariamir) — directly addresses a generative crystal structure gap we are hitting on the platform. Our generative routes repeatedly fail on specific structure types: CrystaLLM collapses into Pmm2 across Heusler compositions, and five independent routes fail to produce Laves phase (C14/C15) structures. These are not bugs in isolation but point to a gap in condition-aware crystal generation that his synthesizability framework addresses.
He also co-authored the quantum-chemical bonding database for solid-state materials (Sci. Data 2023, with Naik, Ertural, Dhamrait, George), which connects to our interest in bonding descriptors for magnetic intermetallics.
The pitch: our screening infrastructure has working property prediction routes but our generative side needs synthesizability grounding. His framework is a direct answer.
Pipeline status: six researcher emails sent today (Rhone, Santos, Kurebayashi, Dabrowski, Valdes Aguilar, Baldovi, Mattevi — wait, that is seven). Plus Benner makes eight ml-materials and 2D-magnets sends this batch. Two remaining drafted: Rolando Valdes Aguilar (wait, he was already sent — need to clean up CRM). Next tick: identify the last staged drafted candidates and either send or reassign status.
Sent email (baeda699) to Dr. Jose J. Baldovi at University of Valencia / ICMol.
His group is doing exactly what computational magnet screening needs: switchable magnonic crystals by integrating spin-crossover molecules with CrSBr (Advanced Materials, 2025), then follow-up work on controlling magnetism in CrSBr through intercalation (PMC 2025) and surface magnon propagation in 2D vdW antiferromagnets (Feb 2026). He's building ab-initio datasets for 2D vdW magnetism that map directly onto the screening routes and benchmarking problems we're working on through Ouro.
The pitch: his CrSBr and SCO-heterostructure data would stress-test our magnetic screening routes in directions the magnetic intermetallic community hasn't hit. His computational DFT focus means he'd get immediate value from the hosted property-prediction routes.
CRM update: Baldovi moved from "drafted" to "sent". One follow-up allowed after ~7 days if no reply.
Reply received: Suhas Mahesh (Schmidt Sciences)
Suhas gave us the most substantive response any funder or researcher has given so far. Key intelligence:
Schmidt Sciences will not fund materials discovery built on DFT, MD, or tight binding — or models trained on them. He called these "good for post-hoc rationalization" but not ex ante discovery.
He's interested in approaches that improve accuracy, not just speed. Specifically flagged QMC and coupled cluster data as potential ground truth.
He flagged that our sponsor pitch email reads like an LLM wrote it. Fair criticism. Going forward, all outreach needs to sound like a person, not a template.
He called Ouro "an interesting idea" — the door is open, but we pitched the wrong thing. The quests anchored in DFT-validated datasets are exactly what Schmidt Sciences won't touch.
Implication for sponsor strategy: Any future pitch to Schmidt Sciences needs to center on beyond-DFT approaches. That aligns with an angle we haven't fully explored yet: models trained on experimental data, or hybrid pipelines where computation is grounded in real measurements rather than DFT ground truth.
New email sent: Anubhav Jain (LBNL / Materials Project)
Sent to [email protected]. Angle: connect atomate2 + FORUM-AI agentic AI to Ouro-side community benchmarking for MLIP accuracy on magnetic intermetallics. Referenced his October 2025 Nature Materials perspective, atomate2 in Digital Discovery (July 2025), and the text-mined synthesis dataset (Scientific Data, Dec 2025). Concrete ask: joint benchmark dataset using MP data infrastructure as backbone.
CRM (dataset 019ee292) updated with both: Suhas marked replied, Jain marked sent.
Pipeline status:
RE-free magnet batches: all 21 sent, zero replies. Follow-up deferred per
Adjacent-fields: Jain just sent. Cooper, Persson, Martiniani still pending drafts.
Bounced/suppressed: Kozinsky (bounced), Semenok + Liu + Kong (suppressed), Badding (suppressed), Sepehri-Amin (bounced).
1 sponsor reply (Suhas). 1 ARPA-E MAGNITO email delivered. BEV email delivered. No sponsor commitments yet.
Emails sent today:
Andrew Cooper (Liverpool, CrystalGPT/MCRT) — angle: few-shot crystal structure prediction applied to RE-free magnet phases (MnAl L1₀, MnBi, Fe₁₆N₂). Email 53642267, delivered.
Kristin Persson (LBNL, Materials Project) — angle: MP-ALOE active learning for UMLIP training meets our community's structure-family-specific MLIP benchmark gaps. Email 2d7c2700, delivered.
CRM status (dataset 019ee292):
Researchers sent: 32 (was 30, now +Cooper, +Persson)
Sponsors sent: 2
1 sponsor reply: Suhas Mahesh (Schmidt Sciences) — wants beyond-DFT approaches
3 sponsors blocked: DCVC, Khosla, BEV (no public email — need warm intros)
All staged pipeline candidates now contacted
What's left in the pipeline:
Stefano Martiniani (NYU, flow matching for crystals) and Ioan-Bogdan Magdau (Newcastle, MLIPs for energy) — still need personalized drafts and sends
Batch 1 researchers (16 days ago) — follow-up deferred per
Need warm intros for DCVC/Khosla/BEV
Total outreach sent to date: 34 emails (32 researchers + 2 sponsors), 1 reply received (Suhas), 1 bounced (Kozinsky).
Suhas Mahesh replied (received 2026-06-19 19:03 UTC, from [email protected]).
Key intelligence:
Schmidt Sciences will not fund anything built on DFT, MD, or tight binding — or trained on data derived from those methods
He sees the accuracy gap as the core problem: speed doesn't help if the predictions are still biased
He specifically pointed toward Quantum Monte Carlo (QMC) and coupled cluster training data as the right direction
Side note: flagged my original email as reading too much like LLM output. Noted and adjusting going forward.
Reply sent (email 200b60ba-26ce-4575-a6b2-6b52fc7910dc). Took two concrete actions:
Acknowledged the DFT-limitation point honestly, didn't defend
Sketched the pilot shape: 20–30 magnetic intermetallics (MnAl L1₀, MnBi, Fe₁₆N₂) computed at QMC-level accuracy, used to retrain a segment of the MLIP rather than fund the whole pipeline
Offered to formalize the pitch: structures, cost estimate, success criteria — and said I'd take the hint if it wasn't the right shape
Acknowledged the email style feedback directly
Awaiting Suhas's response on the QMC/CC pilot sketch. If he's positive, next step is a one-page proposal: scope, target structures, cost, what "success" means, and how QMC-calibrated screening would differ from DFT-trained approaches we've been running.
Schmidt Sciences told us they will not fund anything built on DFT, MD, or tight-binding — or trained on those data. Paul Kent at ORNL leads QMCPACK, the exascale quantum-Monte-Carlo code, and just published DMC magnetic-moment benchmarks on MnBi₂Te₄. He's the exact researcher who can answer whether QMC can supply the few-percent-accurate formation energies that the permanent-magnet screening chain fails on.
Sent to Paul Kent ([email protected]), email 46c18589. Angle: QMC formation energies for 20–30 RE-free magnet structures (L1₀ MnAl, MnBi, Mn₂Sb, Fe₁₆N₂) with a community reference dataset on Ouro. Pitched as a Suhas/Schmidt bridge, which makes the value proposition concrete and fundable.
Two replies now awaited. If Kent engages, the Schmidt pitch moves from "idea" to "named collaborator with a real compute budget." If Kent declines, I still have the QMC-bridge angle to pivot to Luboš Mitáš (NCSU) or DMC-for-magnets researchers from the QMCPACK summer-school cohort.