课程大纲
实习项目
Participated in a Phase III RCT evaluating progression-free survival for an immuno-oncology drug, contributing to statistical programming and primary efficacy analysis.
Built ADaM datasets from raw SDTM using SAS 9.4, ensuring CDISC-compliant traceability and alignment with SAP specifications.
Developed Cox proportional hazards models using PROC PHREG and conducted subgroup KM plots via R (survival, ggplot2) for clinical insight.
Produced validated TLFs to support FDA NDA submission, including stratified efficacy listings and interactive safety tables.
Assisted in reviewing the Statistical Analysis Plan, refining censoring rules and sensitivity analysis scenarios based on feedback from senior biostatisticians.
Collaborated with the data management team to resolve queries and support database lock.
Delivered documentation and submission-ready programs via Git-based version control, reducing code review time by 30%.
Presented key results and risk interpretation during internal regulatory meetings, supporting final CSR decision making.
Developed a cost prediction model using Python (pandas, scikit-learn) to analyze Medicare Advantage claims and support premium pricing strategies.
Engineered 150+ features capturing utilization patterns, chronic conditions, and demographics using SQL (Snowflake) and Python.
Trained and tuned GLM, Ridge, and XGBoost models to optimize RMSE, improving R² by 18% over baseline actuarial models.
Used SHAP (Python) to interpret model outputs and provide explainability to pricing and actuarial teams.
Integrated scoring pipelines into Snowflake and deployed batch runs using Airflow + Python operators.
Presented insights with model dashboards built in Tableau, and supported actuarial modeling decisions for the next cycle.
Led statistical analysis for post-market surveillance of a respiratory vaccine, integrating EHR + VAERS data to track safety signals in near real-time.
Cleaned and processed over 10 million records using SQL (PostgreSQL) and pandas, building structured datasets for signal detection.
Applied PRR and EBGM algorithms using custom R scripts to identify adverse event-vaccine pairs with statistically significant elevation.
Built a Shiny dashboard enabling pharmacovigilance teams to drill into event rates by region, age, and comorbidity profiles.
Automated safety metric updates using Airflow + Rscript, reducing manual reporting effort and ensuring consistency in weekly reviews.
Cross-validated internal findings with published studies using meta and forestplot packages to enhance medical credibility.
Supported regulatory reporting by contributing to two safety memorandums submitted to the FDA and EMA.
Generated reproducible analytics pipelines using R Markdown, which were later adopted by two other product safety teams.
Conducted an RWE study using Optum EHR to compare treatment outcomes between two oral diabetes therapies.
Extracted and filtered patient records using SQL + SQLAlchemy (Python) with custom inclusion/exclusion rules.
Applied propensity score matching (PSM) using R (MatchIt, tableone) to minimize bias from treatment selection.
Evaluated endpoints like HbA1c reduction, ER visit rates, and treatment discontinuation using Cox and logistic models.
Created stratified subgroup analyses across age, gender, and comorbidity level using survival and survminer.
Built plots and comparative tables with ggplot2 and exported to LaTeX for medical review.
Results were included in a publication draft submitted to JAMA Open and used by medical affairs to guide promotion strategy.
Demonstrated real-world effectiveness signal that supported expanded payer coverage in two regions.
Performed independent QC programming of efficacy/safety endpoints for solid tumor NDA using SAS Base/Macro + R.
Validated all ADaM datasets using Pinnacle21 and custom R scripts to confirm CDISC compliance.
Built traceability documentation linking derived variables to SAP-specified endpoints with dataset lineage.
Created macros and reusable scripts to standardize review across 40+ TLFs.
Flagged and corrected logic errors in adverse event derivation that would have delayed submission.
Assisted regulatory writing by providing footnote explanations, analysis rationale, and visual outputs.
Delivered clean QC package two weeks ahead of timeline, reducing regulatory audit preparation time by 30%.
汇聚1300+
在职面试官
Senior级、 大厂、 在职面试官 ，而想要加入导师库，只满足这3个条件还远远不够。
如同同学求职一般，我们的导师也会历经多达5轮的“面试考核”，通过背景经历、技术能力、面试经验、导师sense等维度的层层筛选下，入选率仅10%。有导师戏称，“并不比一次大厂面试简单”。
为了保证导师库水准，我们从未因考核成本而降低标准。辅导内容随着科技届趋势不断迭代更新，对导师的培养和考核标准也会只增不减。
全球各大一线大厂
专属导师1对1直播授课- 精准匹配头部公司在职面试官
- 成功签约全职OFFER为止
- 无OFFER承诺退款
- 简历项目双修双审
- 一线大厂List，在职员工内推
- 附赠真实实习项目
直通硅谷成立于2015年3月，由北大计算机系师兄联合MIT、前百度网络科技产品经理、Harvard高级学者、香港上市公司联席董事共同创立，心之所向，是壮大全球华人科技力量。
凭借在科技求职辅导中积累的丰富经验，我们不断研发顺应科技界求职趋势的学练结合课程，组建富有实战经验的国内外名企导师团队，已成功帮助超过8000+学员进入全球头部科技公司。
