Nicholas Jenkins
ndjenkins.comLead Data Scientist at TikTok · ex-BCG · Production AI and evaluation for regulatory reporting
Melbourne, Australia · Australian citizen linkedin.com/in/ndjenkins-au
Experience
TikTok (ByteDance) · Lead Data Scientist, Trust & Safety Transparency
New York · May 2022 – Present
- Built the regulatory metrics platform behind TikTok's transparency reporting: reporting time down 75%, ~45% of disclosures self-served without data science, 150+ external disclosures in 2024.
- Built 16 of 23 metric queries behind TikTok's first EU DSA transparency report, published three days early.
- Designed the quality checks behind our independently audited reports, then evolved them into an agentic launch gate: 90% hold recall in backtest and 90% less manual effort.
- Built an in-house LLM inference platform from scratch, load-balanced across up to 300 preemptible GPUs with automatic retries: 4x fleet throughput, 10M inferences.
- Created a statistically weighted LLM insights product used for seven national elections, breaking-news crises and AI scam classification for TikTok's Frauds and Scams Prevention Hub.
- Built a support-ticket AI classifier with rulebook retrieval (RAG), A/B-tested model versions and an evaluator against QA's 23,000-label golden set.
- Built the data contract behind TikTok's disclosure intake; request handling fell from 182 to 15 hours.
- Designed T&S AI-adoption measurement; founded my pillar's 1,000+ PR monorepo and set its code and AI standards (approval-before-merge 37% → 80%); mentored five data scientists; seeded a 49-skill agent library.
Boston Consulting Group · Senior Data Scientist
New York · Jul 2019 – May 2022
- Optimized a B2B distributor's $1B promotional spend through SKU-level price-elasticity modeling.
- Built a $52M personalization pilot for a fashion retailer, lifting sales about 25% above baseline.
- Built an explainable pricing model for a payroll provider, cutting pricing cycles and churn.
- Led a five-person team building BCG's internal feature store, doubling its data assets from 20 to 40.
Evolve Research · Head of Data Science
Melbourne · Feb 2014 – Apr 2019
- Led analysts and engineers delivering analytics for a national broadband provider, supermarket and insurer.
- Built a text-classification and sentiment capability that saved a client $30K a year and sold for up to $300K.
- Built an NLP survey chatbot that lifted respondent engagement and insight quality.
Earlier · Research Associate, Klein Partnership (2011–14) · Account Coordinator, Ross Partnership, advertising (2010)
Education
MS Computer Science, Machine Learning · Georgia Institute of Technology, 2019 · GPA 3.9
BA Psychology & Political Science · University of Melbourne, 2009
Skills
AI & LLMs LLM evaluation, golden sets, A/B model versioning, RAG, agentic workflows, open-weight serving (Ollama, Llama, Gemma), GPU inference, context and prompt engineering, Claude Code
Production Python, SQL, PySpark, Docker, AWS, Git, CI/CD, data observability, drift monitoring, backtesting
ML & statistics scikit-learn, TensorFlow, classification, time series, stratified sampling, experimental design
Leadership technical leadership, architecture, code review, mentoring, regulatory stakeholders (DSA)