Jize Luo Email & Phone Number

United States

Los Angeles, California, United States

[Feb-Jun] Report directly to Head of Data- Leasing Dashboard, Prospect Scoring, Data Pipeline.[Jun-Now] Report directly to Director of AI- AI leasing agent evaluation & prompt strategy, AI for Business Intelligence, SOP knowledge base.

Los Angeles, California, United States

• Extract and compile job requirements from state-level and US federal websites to create a comprehensive database that highlights key competencies and qualifications needed across different government roles
• Systematically evaluate job specifications across government agencies and identify roles most at risk of AI-induced changes and those necessitating new skills to provide targeted recommendations for workforce.
• Assign a five-year impact level to each occupation based on the transformative effects of AI on employment patterns, skill requirements, and occupational resilience within the federal sector to prioritize resource.

Los Angeles, California, United States

• Collaborate with the team to synthesize a diverse dataset by generating 3,000 varied examples of "bad writing" to improve model robustness and accuracy in distinguishing between high and low-quality text
• Leverage the CARC computing cluster to fine-tune the LLAMA-2 model, optimizing it to align with established writing standards and enhancing performance through key metrics such as BLEU Score and Macro F-1 Score
• Identify and systematically address eight domain-specific areas within the writing model, implementing targeted refinements to ensure the model's outputs consistently adhere to the specific conventions and expectations.

Shanghai, China

• Deploy personalized chatbots by fine-tuning large language models, such as ChatGLM-6B via AdaLoRA, to enhance customer interaction and achieve cost reductions, evidenced by a 60 person-day cost saving for BMW's product.
• Develop and maintain data pipelines for extensive driving datasets, incorporating advanced analytical methods like time series causal inference and random forest classification, with a notable project handling.
• Perform driver segmentation using K-means clustering and enhance the model by improving data quality, to facilitate product decisions
• Design and implement innovative data visualizations, such as hexagonal maps, using synthetic feature generation to create engaging and informative trip summaries, significantly enhancing user engagement and awareness.

Berkeley, California, United States

• Transform extensive voting data into structured panel datasets, using 32 years of precinct-level data from the California General Data Bank
• Develop and maintain robust data pipelines to streamline the process of data acquisition, cleaning, and aggregation, showcasing efficiency in handling approximately 200.csv files
• Develop algorithms to adjust and match precinct-level data via geographical shape files
• Conduct advanced text analytics on suffrage proposals, employing tools like NLTK for deep analysis, including custom tokenization and frequency distribution to extract key insights on resident welfare and housing.

Beijing, China

• Optimize SQL queries for metrics such as DAU, acquisition rate, and session time from 3 million entries, achieving a significant reduction in run time by approximately 20% through enhanced sub-queries
• Perform comprehensive funnel analysis to identify user drop-off points across 30 interactive elements, utilizing instrumentation log data and presenting findings through Tableau visualizations to aid cross-functional.
• Design and execute a four-week A/B testing framework for the "Rewards Redemption" module, leveraging guardrail and weighted DAU metrics to assess feature enhancements and their impact on user engagement
• Refine prediction models for membership subscriptions and establish an inference pipeline using Amazon SageMaker, addressing class imbalance with SMOTE to improve the macro F-1 score from 89% to 95%, thereby aiding the.
• Implement DiD (Difference-in-Differences) analysis to optimize the "Rewards Redemption" module's prize pool, analyzing customer session times and redemption behaviors to reduce operational expenses by 20%

Master Of Science - Ms, Applied Data Science

Bachelor Of Arts - Ba, Statistics & Economics, 3.85