What @sandeepnailwal said about AI hit me hard.
“AI is a centralizing force. The only way to protect humanity is to make it open and accessible to everyone.”
He compared it to how nuclear deterrence worked , balance came when power was shared, not controlled.
That’s exactly why I’m bullish on Sentient.
It’s not just about building smarter models , it’s about giving everyone the power to protect themselves in the age of AI.
The risk of having a world where AGI is closed and wielded by a few big centralized entities is just too high—@sandeepnailwal’s vision challenges this.
The hypothesis: building a collaborative foundation of open source AI tech stack that accelerates progress toward secure, and composable high-level reasoning intelligence, will foster a future where AGI serves and enhances humanity.
This is Sentient.
@SentientAGI@sandeepnailwal Exactly. Open AGI isn’t just a tech choice , it’s a moral one. Transparency is how we make sure intelligence serves everyone.
The research behind @SentientAGI is on another level.
This isn’t just an AI startup , it’s a team that just had 4 papers accepted at @NeurIPSConf, led by researchers from Princeton, UIUC, DeepMind, Meta, and Stanford.
At the core is Pramod Viswanath, cofounder of Sentient and Professor at Princeton.
He helped create OML (Open Model License) , a framework that lets open models be owned, traced, and verified without losing transparency or performance.
This work basically gives open-source AI a legal and technical backbone.
Sala, a former DeepMind and Meta researcher, leads Sentient’s work on multi-agent systems , how AIs can cooperate, plan, and solve complex tasks safely.
He helped develop ROMA, Sentient’s framework that lets agents coordinate and execute across different tools.
Edoardo, who studied Physics at Princeton, connects deep theory with real AI systems , studying how Transformer models behave and how to make them more robust and understandable.
And Peiyao, who worked with Chainlink Labs, focuses on the security and governance side , ensuring AI agents and networks are cryptographically verifiable and resistant to attacks.
Together, they’re building the science that makes open, safe, and verifiable AI real , not just talk.
FutureTech @SentientAGI
Our research team, drawn from top universities and leading tech companies, has developed cutting edge technology across models, agents, reasoning, benchmarks, and AI security. Here’s a look at the standout contributors behind these advances.
This is the team that achieved four papers accepted at @NeurIPSConf across multiple tracks.
@viswanathpramod , cofounder of Sentient and Professor of Electrical & Computer Engineering at Princeton University, bridges information theory, cryptography, and learning theory to make open models verifiable, attributable, and controllable.
He co developed the OML paradigm defining ownership, control, and alignment for open models and co led OML 1.0 (Fingerprinting), which proved that large scale model attribution can preserve both performance and transparency.
A former UIUC professor, Viswanath is renowned for translating deep theoretical advances into practical, open-source AI frameworks. His work also integrates cryptographic accountability, licensing economics, and governance into a unified, fiduciary first foundation for the open model ecosystem.
@sala88232 , one of Sentient’s core researchers on multi agent systems, holds an MS in Computer Science from UMass Amherst and has conducted research at Google DeepMind, Meta FAIR, and Microsoft Research.
His work spans both academia and industry, focusing on the generalization capabilities of machine learning models, especially large language models (LLMs). He coauthored FLAMe, a family of foundational autoraters for reliable automatic evaluation, and has led Sentient’s efforts in Open Deep Search and ROMA.
At Sentient, he explores multi agent coordination for long-horizon tasks, pioneering recursive control frameworks within ROMA atomize, plan, execute, aggregate to enhance reliability, sample efficiency, and cross tool alignment on complex queries.
@edoardocontente , one of Sentient’s earliest researchers, earned his bachelor’s degree in Physics from Princeton University, where his senior thesis extended the theoretical framework of deep learning to study the initialization behavior and implicit bias of Transformer models. His work provided a modular and perturbative analysis linking general deep learning theory to the specific dynamics of modern Transformer architectures.
He went on to complete his Master’s in Electrical & Computer Engineering at Princeton, before joining Sentient, where he has contributed across multiple research fronts including OML 1.0, Open Deep Search, and most recently, a study on the robustness of model fingerprints.
@peiyaosheng , one of Sentient’s earliest researchers, earned her PhD in Computer Science at UIUC under Pramod Viswanath, was a visiting researcher at Princeton, and previously studied in the ACM Honors Class at Shanghai Jiao Tong University.
Her work sits at the intersection of decentralized systems and AI agents, focusing on building secure, governable, and measurable open networks and stress testing how agents behave atop those foundations.
At Sentient, Peiyao contributed to OML 1.0 (Fingerprinting) and evaluation projects like LiveCodeBench Pro, while advancing agent-security research (such as fatal context manipulation attacks on online/Web3 agents) and governance reasoning.
Her broader research spans cryptoeconomic security and forensics for distributed ledgers including Proof of Diligence for rollups, BFT protocol forensics, TrustBoost for cross chain trust, ACeD oracles for data availability, safe light clients, proofs of backhaul for bandwidth attestation (NDSS ’24), and PoLoc proofs of location.
Beyond academia, she co-founded Witness Chain to unify the DePIN economy, and brings industry experience from Chainlink Labs, where she worked with Dahlia Malkhi. Across her papers and systems, her guiding goal remains constant: to build cryptographically verifiable and auditable foundations that allow open models and agents to operate safely at scale.
@sewoong79, Director of AI Research at Sentient and Professor at the University of Washington’s Paul G. Allen School (formerly at UIUC), leads Sentient’s research across the AI stack.
After earning his PhD in Electrical Engineering from Stanford, he focused on differential privacy, federated learning, robustness, and optimization, including experience at Google. He now applies rigorous theory covering generalization under distribution shift, private and robust optimization, and online learning to production-grade systems, ensuring ownership proofs and behavior constraints remain intact through fine-tuning, distillation, and tool-augmented execution.
Through his scholarship and leadership, Sewoong drives verifiable attribution, safe model control, and practical privacy, translating frontier research into deployed capabilities across Sentient’s AI stack.
@socrates1024 is a coauthor of the OML paper and brings deep expertise in cryptography and confidential computing. He works at Teleport, focusing on confidential compute applications for social media and communications, holds a PhD from the University of Maryland Cybersecurity Center, serves as an Adjunct Associate Professor at UIUC, and is Associate Director of IC3 (@initc3org ).
At Sentient, Andrew contributed to the confidential compute and TEE mechanisms within OML, supporting frameworks that make open-source AI artifacts open, monetizable, and compliant.
@SentientAGI@vivekkolli@abhishek095@sandeepnailwal@hstyagi@0xsachi@antoniok1406@oleg_golev @scalpkripto