Design & UX

Optimize where multimodal models attend by treating attention weights as a learnable policy, using policy gradients with advantage weighting to improve visual grounding and perception without changing model architecture.

Transform lengthy documents into fully narrated presentation videos with synchronized audio-visual delivery. Automatically segments content, generates visuals, synthesizes speech, and composes final video.

Generate consistent ultra-long videos (up to one minute) using depth and trajectory controls with autoregressive segment generation and unified noise initialization.

Convert webpage designs to code via Layout-as-Thought reasoning, detecting layout structure and generating HTML/CSS for spatial blocks.

Train robot controllers via five-stage curriculum progressing from base vision-language models to embodiment-specific RL-refined policies. Unified action space enables cross-embodiment transfer with minimal performance loss.

Systematically adapt state-of-the-art distillation methods for T2I generation. Compare sCM (stabilized Consistency Models), MeanFlow, and IMM within unified framework. sCM excels at extreme few-step regimes (52.81% GenEval at 2 steps), MeanFlow achieves superior fidelity at 4 NFEs.

Convert video generation model outputs into executable robotic manipulation by extracting 3D object flow trajectories as an intermediate representation. Enables zero-shot manipulation of diverse object types (rigid, articulated, deformable, granular) without task-specific training. Use when pre-trained video models capture plausible manipulation patterns but need grounding in low-level robot control.

Enable Diffusion Language Models to achieve 8.5x inference speedup (24-30 steps vs. 256) through certainty-forcing distillation that trains models to achieve simultaneous high confidence across multiple tokens. Use when optimizing inference latency for dLLM deployments.

Generate physically grounded 4D scenes from natural language through dual-stream architecture separating object detail from scene orchestration. VLM-Motion Critic validates physics parameters iteratively, bridging semantic-physical execution gap.

Generate images with unified control over identity, spatial position, pose, and layout: encode diverse control modalities (spatial canvas, pose canvas, box canvas) into single RGB image, train diffusion model jointly across all control types, and enable flexible multi-modal composition at inference without task-specific fine-tuning.

Automates artistic typography customization through self-distilled learning and localized style injection. Generates stylized text images by encoding reference style and injecting it into diffusion denoising. Use for digital design workflows, text-based visual content creation, or applications needing artistic typography control without manual annotation.

[Code Quality] Use when reviewing UI code for accessibility, responsiveness, performance, and UX best practices.

[Code Quality] Use when reviewing in-scope user interfaces for content fit, supported-size layout, project styling conventions, layering where applicable, accessibility, and async states.

[Architecture] Use when scaffolding project foundations and golden-path examples selected by the target project architecture before feature implementation.

[Planning] Use when creating an implementation plan. Flag: --mode={ci|cro} (default standard); ci plans a fix from a CI run, cro plans conversion-rate optimization.

[Implementation] Use when coding and testing an existing plan. Flags: --approval=off, --tests=off, --parallel={auto|on|off} (default off).

[Project Management] Use when generating an HTML mockup report from PBI and story artifacts.

[Documentation] Use when synthesizing specs, PBIs, ideas, and mockups into one standalone HTML slide deck for stakeholders.

[Implementation] Use when implementing a feature step by step.

[Design] Use when creating or describing a UI design. Flags: --mode={fast|good|describe|screenshot|video} (default fast), --lane={product|marketing} (default product).

[Project Management] Use when creating UI/UX design specs from requirements, PBIs, or stories. Flag: --mode=wireframe converts sketches into structured specs.

[Code Quality] Use when reviewing UI code for accessibility, responsiveness, performance, and UX best practices.

[Planning] Use when creating an implementation plan. Flag: --mode={ci|cro} (default standard); ci plans a fix from a CI run, cro plans conversion-rate optimization.

[Project Management] Use when generating an HTML mockup report from PBI and story artifacts.