ChiChieh Huang

Rebuilding how humans absorb, connect, and evolve knowledge. Creating a world where knowledge organizes itself and collective wisdom can emerge. Website：https://ailogora.com/

• Delivered multiple presentations on the latest developments in Generative AI. • Provided consultation and strategic planning for generative AI product pipelines. • Developed a text-to-video product integrating text-to-speech and Avatar technologies. • Engineered and optimized a Retrieval-Augmented Generation (RAG) system using LangChain, including PDF optimization and memory management. • Assisted clients in automating the migration of on-premises applications to the cloud, primarily utilizing AWS.

🏆 GenAI Stars Competition – Top 20 Led the team to become one of the top 20 finalists out of 263 teams in the GenAI Stars Innovation and Entrepreneurship Competition, recognized for innovation and execution in AI-powered storytelling. 🔧 GenAI Technology Development Spearheaded the research, verification, and implementation of Generative AI technologies, including LLM integration, AI avatar generation, and Mandarin voice cloning systems. Ensured both innovation and technical robustness across key product components.

Driving multimodal GenAI product development across TTS and real-time avatar animation systems. • Led the development of personalized TTS voice cloning pipelines, integrating cutting-edge models such as SoVITS, RVC, and Bark to generate natural, expressive synthetic voices • Built and deployed real-time avatar systems combining lip-sync, facial animation (Audio2Face, SadTalker), and character rendering, fully integrated into a web-based experience • Architected robust inference infrastructure, implementing GPU resource locking, failover recovery, and task orchestration for high concurrency and stability • Developed scalable AI backend services with Flask, gRPC, and Celery, managing distributed tasks and integrating GenAI APIs for TTS, deepfake, and story generation pipelines • Collaborated closely with product and design teams to deliver emotionally engaging storytelling experiences through voice-driven avatars

• The Product Owner for Causal AI - Led a 6-person team to develop an MLOps product for Casual AI algorithms, securing contracts with 2 clients within a year. - Collaborated on developing a DL-based Causal Discovery algorithm to autonomously identify causal relationships in data, integrating SEM to evaluate the stability of causal graph structures and providing a visual causal graph product. - Initially responsible for MVP to assess market demand and backend API setup; mid-term collaborated on product architecture and dashboard design; later phase focused on algorithm development and business modeling to establish a clear product concept. • Project lead Collaborated with clients to define objectives, clean data, build models, and deliver insights. 1. Top-three global pharmaceutical company:  Developed an MMM model using Random Forest, optimizing marketing strategy contributions, and budget allocation, saving the client 20% in expenses while boosting revenue by 5% with precision targeting through Uplift Modeling. 2. Leading advertising firm in Taiwan: Created channel-specific MMM models with CatBoost to enhance digital ad spending efficiency, resulting in a 10% revenue increase. 3. Major global steel and aerospace company:  Implemented autotuning with backtesting for various ML models including RF and DeepAR, achieving a 95% forecasting accuracy and a 52% net revenue increase. 4. Top-five global pharmaceutical company:  Developed a questionnaire survey report generation system using the Llama2 language model integrated with RAG, resulting in an 80% time-saving for users. • Commercialization Strategy for R Language Products - Providing a comprehensive commercialization process for R language services. - Implementing encryption and decryption for R language.

• Provided statistical consultations for 6 diverse projects, including survival analysis, genetics, experimental design, and network integration analysis. • Addressed gene selection issues using regression, combining multivariate and lasso techniques, and developed a novel method to create a model, leading to successful submissions. • Self-learned the latest techniques in survival analysis and Hasse diagram from literature reports, aiding consultants in publishing their work.

• Made course materials and led practicums of Applied Biostatistics and R implementation • Build a website to provide students with a systematic approach to learning statistics and programming • Met with students during office hours to answer statistical questions and offered support with R and python.

I contributed to the creation of spatial information charts and conducted in-depth research on the correlation between dengue fever and green spaces, which led to the publication of our findings in an international journal. The project was titled: 'Collaborative Systems for Research Clusters in Mosquito-Borne Disease Control.

I assisted in collecting air pollution data using precision instruments and explored its relationship with green space information in an end-of-term project. The project was titled: "Integrated Multi-Dimensional High-Resolution Information for Assessing Vulnerability to Heatwaves in Taiwan."

- Proposed a statistical methodology research called "Applying structural equation modeling to meta-analysis of diagnostic test performance"