Integrating ACE++ and UNO into a production generative AI platform

The challenge

Both integrations sat deep in the platform’s existing architecture. The work demanded an ML engineer who could read an unfamiliar codebase, evaluate open-source implementations and research for ACE++ and UNO, and ship both capabilities without destabilizing a product professionals already used daily.

The work

• Placed in six days. Eventum placed a senior ML engineer with generative-AI and model-integration experience within a week of the request.
• Repository analysis first. The engineer assessed the platform’s architecture and the open-source landscape for both algorithms, and planned integration points before touching production code.
• ACE++ integration. Subject-driven image generation and editing from a single reference image, added to the platform’s existing generation flows.
• UNO integration. Controllable in-context generation with high consistency across both single-subject and multi-subject use.

The result

Both capabilities shipped modularly on the existing infrastructure: no rebuild, no compatibility breaks, and a platform that could now offer subject-driven and controllable generation to its users. The client got two research-grade capabilities in production from one placement.

Why it mattered

Model integration is its own skill. Papers and repositories get a capability to a demo; putting it inside a live product, behind real traffic, without regressions, is the part that needs senior engineering. That is the gap this placement closed, in days rather than a hiring cycle.

Conclusion

One senior ML engineer, placed in six days, took ACE++ and UNO from open-source research to production features on a working creativity platform.