I combine product judgment with hands-on AI Coding to ship MVPs

Desktop Agent for student job seekers, covering job search, interview practice, resume building, and project training

Role: Product Definition · UI Prototyping · Desktop Full-Stack Development

Jun 2026 - Present

Productized mentor-led project training and resume-ready delivery into an AI workflow, combining a role-training bootcamp mechanism with job-search skills across the path from preparation to offer.

• Product definition: combines real project training with job-search execution to create verifiable portfolios
• Product architecture: role workflow training plus skills for search, interview practice, mock interviews, and review
• Solo build: designed the UI prototype and developed the desktop product with AI Coding

A delivery-oriented desktop Agent for professionals, students, and creators that turns AI capabilities into usable files

Role: Product Definition · Scenario System · Interaction Design · AI Coding Prototype

Jun 2026 - Present

Reframed a developer-oriented Agent runtime as a desktop workspace for mainstream users: choose a scenario, upload source material, describe the task, follow execution progress, and receive editable PPT, Excel, and Word deliverables.

• Product positioning: focused on work, content creation, and job-search scenarios while hiding terminals, code, and model settings behind a Task → Progress → Deliverable flow
• Scenario and Skill system: designed 8 launch-critical Skills covering presentations, Excel analysis, weekly reports, meeting notes, social content, multi-platform rewriting, video scripts, and resume tailoring
• Desktop execution: built the workspace, history, Skill library, file attachments, progress states, and artifact management with Tauri and React while reusing Codex Core
• Presentation delivery lane: designed a structured TaskSpec → DeckSpec → Python rendering and automated QA pipeline for page constraints, grounded facts, editable charts, and reopen validation

A music product concept centered on emotional encounters, combining listening, social discovery, AI music generation, and interaction

Role: Product Strategy · User Insight · AI Music Creation · Feature Design

Early undergraduate project

A concept exploring how a music product can extend beyond playback into content, social discovery, and creation through Familiar Mode, Music Square, Instrument Play, and hands-on AI-generated music and songwriting.

• Product positioning: reorganized listening, discovery, and social interaction around emotional moments in music
• Core concepts: recommendation feed, music square, instrument interaction, AI songs, and virtual identity
• AI music creation: generated original songs around user keywords and emotions, then embedded them into the product experience
• Complete proposal: covered user insight, competitor research, feature design, and monetization across 42 pages

HUST undergraduate thesis — collaborative model training without centralizing user data, balancing performance and privacy

Role: Core Researcher

Jun 2024

Built on Shen et al. (2022)'s Federated Functional Gradient Boosting (FFGB) algorithm, extending it with a privacy protection mechanism. Clients train weak learners via residual tracking each round; the server aggregates them into a global model. An adaptive Gaussian noise mechanism dynamically adjusts noise intensity based on loss changes to balance privacy and accuracy.

• Integrated Boosting into the federated framework with an adaptive Gaussian differential privacy mechanism that tunes noise strength dynamically
• Applied knowledge distillation to compress the ensemble model and reduce client–server communication overhead
• Validated across multiple datasets under high data heterogeneity: significantly faster convergence than FedAvg, with strong model accuracy retained after applying differential privacy