Mirrored from filpgf.io — ProPGF Batch 3 (Karma program 1479, application 6a31717e13ef2aeed7ce0d8c, status: pending). Contact details redacted; canonical application lives on filpgf.io. 1.1 Project Name QMesh Vault: Agentic Drug Discovery on Filecoin 1.2 Project Github https://github.com/QMeshPy/distributed-quantum 1.3 Project Website https://distributed-quantum.com 1.4 Team Lead/Point of Contact Soham Bhoir — Founder & Project Lead, QMesh / Distributed Quantum. Preferred communication: Email; Telegram @wintersoren; Discord winter_soren. 1.5 Category [ "RFP 3 - AI infrastructure products on Filecoin" ] 1.6 Open Source Status Fully Open Source 2.1 Project Summary QMesh Vault will make Filecoin the verifiable memory, provenance, and archival backbone for QMesh, an agentic AI platform for automated drug discovery over distributed quantum and classical compute. The project will let biology labs package each discovery run — molecular inputs, workflow graph, agent actions, scoring outputs, in-silico trial-simulation reports, and reusable agent memory — into encrypted, content-addressed experiment capsules stored through Filecoin paid deals. This supports application builders, application users, storage providers, and Filecoin ecosystem pods by creating a customer-facing AI/science product that drives real storage demand. The gap we address is that agentic science platforms need durable, tamper-evident storage for auditability, reuse, collaboration, and licensing, but today these artifacts are fragmented across notebooks, cloud buckets, and local storage. Over the grant, we will ship the Filecoin-backed capsule layer, FEVM experiment and agent registries, a verification dashboard, and a reusable agent/workflow exchange connected to live drug-discovery demos. 2.2 Who does this work support? [ "Pods", "Onramps", "Storage Providers", "Application Builders", "Application Users", "Network Infrastructure", "Other" ] 2.3 Total Funding Requested (USD) 200000 2.4 Milestones & Budget [ { "title": "Month 1.5: Experiment Capsule Spec and Generator", "description": "Project schedule: 6-month grant term beginning 08/01/2026 and ending 01/31/2027. This application funds a realistic v1 Filecoin integration for QMesh, not the full long-term platform vision. The grant focuses on four concrete deliverables: encrypted experiment capsules, FEVM registries, verification tooling, and two pilot drug-discovery demos using Filecoin-backed provenance.\n\nMainnet paid deals are included as a target where operationally feasible. The guaranteed deliverables are the open-source capsule/registry/verification system, Calibration deployment, reusable demo workflows, and a mainnet-ready paid-deal path. We will not claim production clinical-trial readiness, regulated biomedical deployment, or a full commercial marketplace during this grant.\n\nThis milestone creates the core Filecoin-native research object used by QMesh: the encrypted experiment capsule. A capsule is the durable record of an agentic computational drug-discovery workflow. It packages molecular inputs, workflow metadata, agent actions, scoring outputs, benchmark results, and in-silico trial-simulation artifacts into a content-addressed object that can be stored and retrieved through Filecoin.\n\nThe goal is to make Filecoin part of the normal lifecycle of a QMesh run. When a discovery workflow finishes, QMesh should produce a verifiable research object with a manifest, CIDs, encryption metadata, provenance fields, and storage instructions. This milestone is intentionally focused on the data model and packaging layer, because later registries and verification tooling depend on this foundation.\n\nSensitive biomedical artifacts will be encrypted before storage. Public metadata will be limited to what is required for reproducibility, verification, and storage coordination.\n\nBudget allocation: approximately $22,000 core engineering, $5,000 capsule specification and documentation, $4,000 packaging/storage tooling, $2,000 test infrastructure, and $2,000 project coordination.", "dueDate": "2026-09-15", "fundingRequested": "35000", "completionCriteria": "- Public experiment-capsule specification published in the repository.\n- JSON/YAML manifest schema for drug-discovery experiment capsules.\n- Working capsule generator integrated with QMesh workflow outputs.\n- Encryption metadata model documented.\n- CAR packaging and CID generation implemented.\n- Capsule validator implemented to check manifest completeness, CID integrity, encryption metadata, and required provenance fields.\n- At least 10 sample capsules created from demo workflows.\n- At least 3 capsule examples covered: molecular screening, binding/property scoring, and in-silico trial-simulation report.\n- Calibration testnet storage demo or mainnet-ready storage workflow documented.\n- Documentation ad …[truncated] 3.1 Impact pathway QMesh Vault targets these exact 2026 Filecoin Network Objectives: Objective 1 — Drive Paid Onchain Deals: direct contribution. QMesh converts computational drug-discovery workflows into Filecoin-backed experiment capsules, reusable agent packages, and workflow artifacts. The measurable KPI is QMesh-attributable paid Filecoin storage deals. By the end of the grant, the realistic target is 5–10 paid mainnet deals if provider onboarding is available, with Calibration evidence and a mainnet-ready deal path as fallback. Objective 2 — Strengthen Network Profitability & Cryptoeconomics: indirect contribution. QMesh does not modify Filecoin tokenomics, burn mechanisms, lock mechanisms, or protocol economics. Its contribution is demand-side: recurring scientific storage and retrieval workloads for storage providers. Objective 3 — Scale Paid Onchain Flagship Client Adoption: direct contribution. QMesh is a customer-facing AI/science product for computational drug discovery. The measurable KPIs are stored scientific artifacts, registered experiment/agent manifests, retrieval checks, reusable agents, and external pilot users. Output: during the 6-month grant term from 08/01/2026 to 01/31/2027, QMesh will ship a v1 capsule layer, FEVM ExperimentRegistry and AgentRegistry, verification CLI/dashboard, two non-clinical pilot demos, and a mainnet-ready paid-deal path. Outcome: users can create encrypted experiment capsules, register metadata, verify provenance, and reuse basic agent/workflow packages. Impact: Filecoin becomes the practical storage and provenance layer for a real AI/science workflow, with measurable progress toward paid deals and flagship-client adoption. 3.2 Verification metrics Metric: QMesh-attributable paid Filecoin storage deals Network Objective: Objective 1 — Drive Paid Onchain Deals Data source: Filecoin chain explorers, deal IDs, QMesh provenance dashboard How measured: Count unique paid mainnet deal IDs associated with QMesh experiment capsules, agent packages, or workflow artifacts Target by end of grant: 5–10 paid mainnet deals if provider onboarding is available; fallback is Calibration evidence plus documented mainnet-ready paid-deal path Metric: Scientific artifacts stored or prepared for Filecoin-backed storage Network Objective: Objective 1 — Drive Paid Onchain Deals; Objective 3 — Scale Paid Onchain Flagship Client Adoption Data source: Capsule manifests, CAR files, dashboard records, deal metadata where available How measured: Aggregate size of encrypted experiment capsules, agent packages, workflow artifacts, and demo outputs Target by end of grant: 250 GiB+ stored or prepared for paid Filecoin storage Metric: Experiment and agent manifests registered Network Objective: Objective 3 — Scale Paid Onchain Flagship Client Adoption Data source: FEVM registry events, Filecoin/FEVM explorers, QMesh dashboard How measured: Count unique manifest CIDs registered through ExperimentRegistry and AgentRegistry Target by end of grant: 75+ registered manifests Metric: Retrieval verification checks Network Objective: Objective 3 — Scale Paid Onchain Flagship Client Adoption Data source: QMesh verification service, retrieval logs, public demo reports How measured: Scheduled or manual retrieval checks against stored capsule and agent CIDs Target by end of grant: 100+ retrieval checks, with results visible in the dashboard or logs Metric: Reusable drug-discovery agent packages Network Objective: Objective 3 — Scale Paid Onchain Flagship Client Adoption Data source: AgentRegistry events, repository examples, QMesh dashboard How measured: Count agent packages with manifests, CIDs, docs, versions, and demo outputs Target by end of grant: 3 reusable agent packages Metric: End-to-end non-clinical pilot demos Network Objective: Objective 3 — Scale Paid Onchain Flagship Client Adoption Data source: Public repository, demo video, provenance dashboard, registered CIDs How measured: Count complete workflows from computational objective to Filecoin-backed capsule and verification record Target by end of grant: 2 complete demos Metric: External testers or pilot reviewers Network Objective: Objective 3 — Scale Paid Onchain Flagship Client Adoption Data source: GitHub issues, pilot notes, Discord/Telegram records, dashboard activity, written feedback How measured: Count external technical users, DeSci builders, or computational biology reviewers who test or review QMesh workflows Target by end of grant: 2+ external testers/reviewers 3.3 References Manusheel Gupta — Filecoin / FHE ecosystem mentor and QMesh core team member; advised on ProPGF Batch 3 infrastructure alignment, py-libp2p × Filecoin interoperability, decentralized science use cases, and product positioning. GitHub: https://github.com/seetadev Paul Robinson (pacrob) — py-libp2p maintainer and reviewer; can vouch for maintainer contributions, interoperability work, and Filecoin/libp2p networking context. GitHub: https://github.com/pacrob Johanna Moran — libp2p/Filecoin ecosystem adviser; can speak to sustained maintainer work across py-libp2p and py-multiformats, and relevance to Filecoin networking infrastructure. Discord: j0hmo 4.1 Monthly Operating Burn [ "$10-$100K (small team)" ] 4.2 What % of total team monthly burn depends on this grant? Approximately 85–90% of the Filecoin-focused 6-month workstream depends on this grant. The requested $200,000 funds the guaranteed v1 deliverables: capsule spec/generator/validator, Calibration storage demo, FEVM ExperimentRegistry and AgentRegistry, verification CLI, minimal provenance dashboard, two pilot demos, three reusable agent packages, and a mainnet-ready paid-deal path. Additional non-grant support will come from founder overage, Manusheel’s ecosystem strategy support, pro bono ecosystem guidance, existing infrastructure, and community testing. These non-grant contributions improve execution quality but are not required for the core guaranteed deliverables. Contingent/stretch outcomes are limited to mainnet paid-deal volume and larger storage volume, because those depend on storage-provider onboarding and operational coordination. 4.3 If this grant is not awarded, what happens? QMesh will continue as an open-source project, but the Filecoin integration will proceed much more slowly and likely remain limited to documentation, prototypes, and part-time testnet experiments. Without this grant, we would defer the production-grade experiment-capsule layer, FEVM registries, provenance explorer, and agent/workflow exchange. The distributed quantum and agentic drug-discovery platform would continue, but it would not generate meaningful paid Filecoin storage demand in the near term. We would also have less capacity to package reusable drug-discovery agents, onboard pilot labs, run mainnet-ready storage workflows, and build public verification infrastructure. In practical terms, without this grant the project would remain research-driven rather than becoming a Filecoin-native customer-facing product within the Batch 3 horizon. 4.4 Core Team Soham Bhoir — Founder and Project Lead, QMesh / Distributed Quantum. Soham leads product direction, platform architecture, distributed quantum compute coordination, agentic AI workflow design, drug-discovery roadmap, documentation, demo delivery, and pilot-user onboarding, FEVM/solidity engineer. He has founded EUCLID lab a blockchain x quantum lab at his University and was praised by Cyber-Peace foundation, L&T infotech for the design, architecture and leadership. GitHub: github.com/winter-soren Expected allocation: approximately 70–80 hours/week during the grant period. Manusheel Gupta — Core Team Member, Filecoin/FHE Ecosystem Maintainer of py-libp2p. Manusheel supports Filecoin ecosystem alignment, decentralized infrastructure strategy. He brings experience across the Filecoin and FHE ecosystems and will help shape the Filecoin-native architecture, pilot engagement, impact metrics, and long-term sustainability strategy. GitHub: https://github.com/seetadev Expected allocation: approximately 30–35 hours/week during the grant period, focused on strategy, ecosystem alignment, review, partnerships, and ProPGF delivery support. 4.5 Has your team received a ProPGF grant or funding from PLFIF before? [ "No" ] 5.1 Key risks & dependencies Risk 1: User experience complexity. Filecoin storage, CIDs, deals, registries, and retrieval checks can be difficult for biology labs to understand. We will mitigate this by hiding storage complexity behind the QMesh experiment-capsule layer. A lab should interact with simple actions such as “finalize experiment,” “publish agent,” “verify capsule,” and “share workflow,” while QMesh handles packaging, CID generation, storage, registry updates, and verification. Risk 2: Privacy and IP sensitivity around biomedical data. Drug-discovery workflows may involve sensitive molecules, targets, private hypotheses, unpublished results, or commercial IP. We will mitigate this by encrypting experiment capsules before storage, keeping sensitive data offchain, and putting only metadata, CIDs, permissions, and verification status in registries. Labs will choose public, private, collaborator-only, or embargoed sharing modes. Risk 3: Mainnet storage workflow timing. Paid mainnet deals may depend on storage-provider onboarding, deal tooling, or operational coordination. We will mitigate this by building against testnet first, documenting the mainnet path early, and designing the system so the same capsule/manifest/deal flow can move from testnet to mainnet without redesign. Risk 4: Pilot adoption. Biology labs may need time to trust a new quantum-enabled agentic platform. We will mitigate this by delivering concrete demos, reusable agents, onboarding guides, and reproducible workflows. The grant focuses on making the product usable enough for early pilots, not relying on large-scale adoption before the core product exists. Risk 5: Agent reliability and scientific safety. Drug-discovery agents can make flawed recommendations if they are treated as unsupervised authorities. We will mitigate this by making agents auditable, sandboxed, provenance-aware, and designed for human review. QMesh will focus on computational discovery and in-silico trial simulation, not clinical decision-making or unsupervised real-world medical use. Risk 6: Repository readiness and reviewer trust. The QMeshPy repository must contain the active codebase, documentation, roadmap, and Filecoin milestones before final review. We will mitigate this by pushing the active project materials, adding a Filecoin architecture section, publishing milestone specs, and making demo artifacts easy to inspect. Risk 7: Scope control. The full QMesh vision is broad: distributed quantum compute, agentic AI, drug discovery, Filecoin provenance, and lab-to-lab agent reuse. For this grant, scope is limited to the Filecoin-backed provenance layer, registries, verification explorer, and reusable agent/workflow exchange. Core drug-discovery and distributed-compute work will be used as the product context, but the grant deliverables are Filecoin-specific and measurable. Anything else you want to share that we didn't ask? Scope discipline for the $200,000 request: this grant funds a v1 Filecoin integration for QMesh. Guaranteed deliverables are the experiment-capsule schema, capsule generator, validator, CAR/CID packaging, Calibration storage demo, FEVM ExperimentRegistry and AgentRegistry, verification CLI, minimal provenance dashboard, two non-clinical pilot demos, three reusable agent packages, and a mainnet-ready paid-deal path. We are intentionally not claiming production clinical-trial readiness, regulated biomedical deployment, a full commercial marketplace, or large-scale lab adoption during this grant. The goal is to prove that Filecoin can serve as the provenance, storage, and verification layer for agentic computational drug-discovery workflows, using reviewable open-source artifacts and externally inspectable metrics. Contributing to Core Infrastructure? N/A — this application is submitted under RFP 3: AI infrastructure products on Filecoin, not the Core Infrastructure category. Objective 1 Direct Objective 2 Indirect Objective 3 Direct Open Source Context QMesh / Distributed Quantum is an open-source project. The core platform, documentation, Filecoin experiment-capsule specification, FEVM registry contracts, verification tooling, and demo workflows developed under this grant will be published in the project repository. The only data that will not be open by default is lab-submitted biomedical data, encrypted experiment capsules, private agent memory, credentials, and any user-controlled research artifacts that a lab chooses to keep private. Those artifacts will still use open formats and Filecoin-backed storage, but access will be controlled by the submitting lab. This approach supports Filecoin’s objectives by creating open infrastructure for paid, verifiable storage while respecting privacy and IP constraints in drug-discovery workflows.