Hanjun Li Email & Phone Number

Los Angeles, California, US

Torrance, California, US

• Architect, build, and launch reliable credit risk models (XGBoost) and pipelines using historic data to drive underwriting decisions, resulting a risk reduction of 20% on testing data and 11% on originated loans
• Spearhead loss forecasting cash flow simulation logics optimization and migration, create monthly batch job on Databricks and reduce runtime from 11+ hours to 2 hours, and increase loss metrics alignment by 14% by.
• Research new features, design and build a dynamic income model based on binary search to further automate income verification process, reduce loan conversion friction by 18%
• Analyze various data to size opportunity and design metrics for new business and product initiatives including performing A/B testing on outbound call lists to improve applicant conversion rate
• Document updates and processes to close knowledge gap among stakeholders for more efficient collaboration, provide model benchmark reports to ensure legal compliance, and closely monitor model performance

Valley Forge, PA, US

• Prototyped the end-to-end cloud-based data pipeline to perform sales email classification on 180K+ unlabeled HTML formatted text data; yielded a model with 79% weighted precision and 80% weighted recall
• Wrangled sales email data stored on AWS and designed an email parser based on BeautifulSoup to extract relevant text data; annotated unlabeled text with 12 pre-defined sales activities using rule-based algorithm and.
• Leveraged supervised / semi-supervised machine learning techniques (LinearSVC, sepCNN) and variations of word / sentence embeddings (word2vec, doc2vec, s-BERT) to iteratively perform model refinement
• Executed deep-dive diagnostic and predictive analytics to identify sales activity; integrated the model output to carry out downstream task, which was to recommend optimal sales activity that generated the most profit

Blackfriars, London, GB

• Support the processing specific requirements of Unilever R&D and develop new data science functionalities in PAE (an in-house data science platform)
• Write R scripts to enable interaction effects in the current PAE modeling algorithm PLS (partial least squares regression) to improve model quality
• Design user interface, integrate the resulting features to R shiny application and embed the modified R shiny application into the in-house data science platform

• Gathered information such as financial statements for more than 120 start-ups and unicorn companies each week
• Analyzed the gathered information by performing some basic R summary statistics, applied first-stage filtering and emailed the companies to seek potential partnership in City Investment Bonds issuance
• Conducted outreach with other team members to communicate with clients and to collect debts and receivables

Hong Kong, HK

• Calculated expected reimbursement amount and verified more than 80 submitted requests on a daily basis to help with the reimbursement process
• Generated expense reports for business travel and entertainment claims
• Collaborated with General Ledger, AP, and AR teams to keep track of the general ledger

Master Of Science - Ms, Data Science

Bachelor Of Science - Bs, Financial Actuarial Mathematics And Statistics

Associate Of Science - As, Mathematics