Karish Grover (✈️ICML’26)

Visiting Researcher @Meta • Ph.D. ML @CarnegieMellon • Amazon & Meta AI Ph.D. Fellow • Ex @amazon @Microsoft @LinkedIn

[Karish Grover (✈️ICML’26) @karishgrover]

Updated arXiv versions + detailed threads on each paper coming in the next few days.
If you're at ICML and want to chat about geometric foundations for graphs and language, graph ML, or reasoning in generative models — DMs are open. See you in Seoul at #ICML2026!

[Karish Grover (✈️ICML’26) @karishgrover]

Thrilled to share that 3/3 of our papers were accepted to #ICML2026!🎉I'll be in Seoul from July 4–12, presenting these works with amazing collaborators from @Meta, @AmazonScience, & @CarnegieMellon.
These papers span the geometry of natural language, control theory for graph self-supervision, and structural aggregation in Mixture-of-Experts. Poster details below — come say hi if you're at @icmlconf!🧵

[Karish Grover (✈️ICML’26) @karishgrover]

📄[3/3] "DAG-MoE: From Simple Mixture to Structural Aggregation in Mixture-of-Experts" (led by Jiarui Feng, presented by me) w/ @AIatMeta
Scaling MoE along a new axis — learning a DAG over expert outputs instead of a plain weighted sum.
🔗Paper: https://t.co/rs6F6BErnb
🗓Wed Jul 8 · 5:00–6:45 PM · Hall A #107
#MoE #LLM

[Karish Grover (✈️ICML’26) @karishgrover]

Update: Spending the summer at @Meta in Menlo Park, CA, continuing my role as a Visiting Researcher @AIatMeta with @YinglongXia and @zimplex4.

I’ll be in the Bay Area for the summer so feel free to reach out if you’d like to talk research.

[Karish Grover (✈️ICML’26) @karishgrover]

@shrey_twr Congratulations!! ✨

[Karish Grover (✈️ICML’26) @karishgrover]

Honored to be named an Amazon AI Ph.D. Fellow (a.y. 2025–27) ✨

Grateful to my advisors at @mldcmu — Christos F. and Geoff G. — for their constant support. We’ll be working on geometric DL for graphs, in close collaboration with the researchers at @amazon.

#AmazonAIFellowship

[Carnegie Mellon University @CarnegieMellon]

Ten CMU doctoral students pursuing AI research will receive support from @amazon through the company's new AI Ph.D. Fellowship Program. The researchers, from @SCSatCMU and @CMUEngineering, are tackling foundational challenges critical to AI innovation.
https://t.co/T5RTSr2o4J

[Karish Grover (✈️ICML’26) @karishgrover]

@DeepCache @Meta @mldcmu @YinglongXia @zimplex4 @AIatMeta Thank you, Deepanjali :)

[Karish Grover (✈️ICML’26) @karishgrover]

Excited to share that I’ve joined @Meta as a Visiting Researcher for the next year, in their Pittsburgh office, alongside my PhD at @mldcmu. I’ll be working on #GeometricDL and #GraphML, with @YinglongXia & @zimplex4, under the @AIatMeta AI Mentorship Program.

[Karish Grover (✈️ICML’26) @karishgrover]

@surmenok @Meta @mldcmu @YinglongXia @zimplex4 @AIatMeta Thank you, Pavel.

[Karish Grover (✈️ICML’26) @karishgrover]

Excited to be in Vancouver this week for #ICML2025!🌍✈️. Presenting our paper on curvature-based graph anomaly detection:

📍 East Exhibition Hall A-B E2900
📅 Tue, 15 July | 4:30 – 7:00 PM PDT
🔗 https://t.co/DJSeA4tE4v

Drop by to chat about #geometricDL and #graphML! 🚀🧩

[Karish Grover (✈️ICML’26) @karishgrover]

New paper alert! 🚀📈 #ICML2025 @icmlconf

Can graph curvature uncover hidden anomalies overlooked by traditional methods?

💡Introducing CurvGAD, a mixed-curvature graph autoencoder designed to detect curvature-based geometric anomalies.

🔗 https://t.co/Vl30ofdYFk

(1/9)

[Karish Grover (✈️ICML’26) @karishgrover]

Thrilled to join @AmazonScience as an Applied Scientist Intern this summer in Santa Clara, CA! I’ll be working on Graph Machine Learning and Geometric DL with @thvasilo (and others). Would love to connect if you’re around the Bay Area! @mldcmu @awscloud #GeometricDL

[Karish Grover (✈️ICML’26) @karishgrover]

@stephenroller This is going to be my bio.

[Karish Grover (✈️ICML’26) @karishgrover]

Work done with my advisors, Geoff Gordon and Christos Faloutsos (@bear5claw), at @mldcmu @CarnegieMellon

🔗 https://t.co/Vl30ofdYFk
📌 Code out soon — stay tuned!

#ICML2025 #GNN #GraphNeuralNetworks #GeometricDL #AnomalyDetection (9/9)

[Karish Grover (✈️ICML’26) @karishgrover]

New paper alert! 🚀📈 #ICML2025 @icmlconf

Can graph curvature uncover hidden anomalies overlooked by traditional methods?

💡Introducing CurvGAD, a mixed-curvature graph autoencoder designed to detect curvature-based geometric anomalies.

🔗 https://t.co/Vl30ofdYFk

(1/9)

[Karish Grover (✈️ICML’26) @karishgrover]

🔥 CurvGAD outperforms SOTA with lift of up to 6.5 in AUC scores for GAD, across 10 real-world datasets.

🎯 Works across both homophilic & heterophilic graphs, with complex topologies & varying curvature distributions. (8/9)