Compoundarxiv's profile picture. What we're researching @compoundvc | DM @mhdempsey @shelbynewsad @0xsmac @mackenziejem to discuss

Compound Researcher

@Compoundarxiv

What we're researching @compoundvc | DM @mhdempsey @shelbynewsad @0xsmac @mackenziejem to discuss

مثبتة

Our public theses

Enter the Compound Thesis Database An experiment with a few of our working theses that we've pulled from our internal database & made public. It's rough, but we hope it allows founders, operators, & investors to better connect with us & help us learn. compound.vc/thesis



Why Is the Smart Grid So Dumb Missing Incentives in Regulatory Policy for an Active Demand Side in the Electricity Sector esig.energy/wp-content/upl…


Pulsed plasma coiled tubing drilling in crystalline rock environment pureadmin.unileoben.ac.at/ws/portalfiles…


Ultra-High Efficiency Hydrogen Production Using Commercial-Type Solid Oxide Electrolysis Cell System papers.ssrn.com/sol3/papers.cf…


Operational study shows airlines can reroute to avoid ice-supersaturated zones that form contrails, using existing flight planning tools. Cuts non-CO₂ forcing with fuel/time trade-offs—feasible for near term deployment. iopscience.iop.org/article/10.108…


Builds a stereochemically paired sulfinyl-aziridine covalent library; finds KL2-236, a MYC-binding degrader targeting IDR C203/D205, causing proteasome-dependent MYC loss; optimized to KL4-219A—pointing to druggable sites in MYC. biorxiv.org/content/10.110…


Compound Researcher أعاد

(New Essay) Sequencing vs. Equal Odds In building companies navigating the intersection of science and engineering, we have repeatedly notice the patterns of two distinct R&D paths that lead to success, and a middle ground that leads to structural poison.

mhdempsey's tweet image. (New Essay) Sequencing vs. Equal Odds

In building companies navigating the intersection of science and engineering, we have repeatedly notice the patterns of two distinct R&D paths that lead to success, and a middle ground that leads to structural poison.
mhdempsey's tweet image. (New Essay) Sequencing vs. Equal Odds

In building companies navigating the intersection of science and engineering, we have repeatedly notice the patterns of two distinct R&D paths that lead to success, and a middle ground that leads to structural poison.
mhdempsey's tweet image. (New Essay) Sequencing vs. Equal Odds

In building companies navigating the intersection of science and engineering, we have repeatedly notice the patterns of two distinct R&D paths that lead to success, and a middle ground that leads to structural poison.

Advance: Plain Transformer on 3D coordinates (no graphs/equivariance) matches equivariant GNN MAE on OMol25 and runs faster. Learns distance-decaying attention; follows scaling laws; enables MD/MLIPs for drug/material R&D. arxiv.org/abs/2510.02259


QUID: 170-system QM benchmark of ligand-pocket motifs with CCSD(T) & QMC agreeing to ~0.5 kcal/mol. Decomposes NCIs, scores DFTs, flags FF/SE gaps—raising binding-energy fidelity for more reliable lead scoring in drug design. nature.com/articles/s4146…


Review maps advances in wrought Ni-superalloys: PM disks 50–55% γ′ for ≈788 °C; IM 30–44% γ′ bridging 718→PM; AUSC candidates for 700–760 °C steam stressing creep/oxidation/weldability. Implications: hotter turbines, AUSC plants. link.springer.com/article/10.100…


The Necessity of AC: DC for Long-Term Operation: 20000 Hours of Operation on a 70-Cell SOEC Stack ecs.confex.com/ecs/sofc2025/m…


GPN-Star: a genomic LM trained on multi-species MSAs that learns nucleotide-level functional constraint, setting SOTA for conservation & variant-effect scoring across human/mouse/fly/worm—improving rare-variant triage and regulatory design. biorxiv.org/content/10.110…


40% TPV efficiency via high-bandgap tandem III-V cells (1.4/1.2 & 1.2/1.0 eV) with back-surface photon recycling at 1,900–2,400 °C—making TPV viable for thermal grid storage (TEGS) and compact, fuel-flexible dispatchable power. nature.com/articles/s4158…


Loading...

Something went wrong.


Something went wrong.