Core: Implement AI-Powered Payment Routing Optimization Engine

[Smartdevs17]

Context

AgenticPay processes payments across Stellar (Soroban) and EVM chains. Currently, the routing engine selects chains based on static configuration rather than real-time conditions, leading to suboptimal fee/speed tradeoffs.

Current Limitation/Problem

The routing system uses hardcoded chain priorities. It cannot dynamically adapt to network congestion, fluctuating gas prices, or historical reliability metrics.

Expected Outcome

An ML-driven routing engine that selects the optimal chain for each payment based on real-time and historical data, minimizing cost and latency while maximizing success rate.

Acceptance Criteria

• Implement feature extraction pipeline collecting gas prices, confirmation times, and success rates per chain
• Train a lightweight ranking model (XGBoost or similar) for chain selection
• Expose routing decision logs with rationale for auditability
• Achieve <50ms routing decision latency at p99
• Support manual override via admin API
• A/B test framework comparing static vs AI routing outcomes

Technical Scope

• backend/src/services/routing/ - new routing engine module
• backend/src/services/routing/ai-router.ts - ML inference wrapper
• backend/src/queues/routing-evaluation.queue.ts - periodic feature collection
• Prisma schema additions for routing metrics table
• Redis sorted sets for real-time chain performance data
• Integration with existing PaymentService and ChainConnector interfaces
• Edge cases: chain outage fallback, stale data handling, model cold start

[Gozirimdev]

@Gozirimdev has applied to work on this issue as part of the Stellar Wave Program's 6th wave.

Hello,
I’ve resolved similar problems in the past and am confident I can implement a clean, efficient fix quickly if I am given the opportunity.

ℹ️ Repo Maintainers: To accept this application, review their application or assign @Gozirimdev to this issue.

[OmniZlatoon]

@OmniZlatoon has applied to work on this issue as part of the Stellar Wave Program's 6th wave.

Though it's left only few hours left to go, I ensure you my efficiency in my work to get the job don3. PLEASE ASSIGN ME.

ℹ️ Repo Maintainers: To accept this application, review their application or assign @OmniZlatoon to this issue.

[Babigdk]

@Babigdk has applied to work on this issue as part of the Stellar Wave Program's 6th wave.

Hi! I'd like to work on this issue.

I'll design and implement an AI-powered routing engine that replaces the current static chain selection with a data-driven model using real-time and historical performance metrics. This will include a feature extraction pipeline, a lightweight ranking model for chain selection, low-latency inference logic, and integration with the existing payment flow, along with audit logging, fallback handling, and an A/B testing framework to compare routing strategies.

ℹ️ Repo Maintainers: To accept this application, review their application or assign @Babigdk to this issue.

[my908-hue]

@my908-hue has applied to work on this issue as part of the Stellar Wave Program's 6th wave.

Hi maintainer,iw will really love to work on this project i will Implement feature extraction pipeline collecting gas prices, confirmation times, and success rates per chain
Train a lightweight ranking model (XGBoost or similar) for chain selection
Expose routing decision logs with rationale for auditability

ℹ️ Repo Maintainers: To accept this application, review their application or assign @my908-hue to this issue.

[drips-wave]

Congratulations, @my908-hue! 🎉 Your application was accepted by the repo's maintainers, and the issue is due on June 30, 2026.

🧑‍💻 @my908-hue: Please resolve the issue such that the repo's maintainers have enough time to review your contribution before the due date. You'll earn Points for completing the issue on-time, which will make you eligible for a share of the Stellar Wave Program's reward pool.

Warning

When opening a PR, please link it to this issue to ensure it gets tracked accurately. Points are awarded when this issue is marked as completed by the maintainer.

🤠 Repo maintainers: Please keep an eye on the contributor's progress and review their work before the due date. You can manage this issue, including adjusting its complexity and points, here.

🌊 Happy Wave 🌊