Data Analyst - Los Angeles, CA
Data analyst with 4+ years of experience. Currently the sole analyst for a 3-entity, 12-location retail group in LA, where I run consolidated reporting across two POS systems covering sales, margin, inventory, and P&L. Before that, financial audit at an LSE-listed company, which is where I learned to assume nothing, validate everything, and document the trail. I build dashboards, write SQL across messy multi-source data, and build AI-augmented tools with LLM APIs. My portfolio extends into credit risk, fraud, insurance, and time-series forecasting.
Live demo: AI Resume-JD Matcher - Streamlit app using Claude API + semantic search
Authorized to work in the US, no sponsorship needed. Open to Data Analyst roles (LA or remote), available to start immediately.
LinkedIn · nurbol.sultanov@gmail.com · Tableau Public
Languages: SQL (PostgreSQL, T-SQL), Python (pandas, NumPy, scikit-learn, SciPy, statsmodels, Plotly) BI & Visualization: Tableau (live), Power BI (DAX, Power Query), Excel (advanced) ML & Statistics: Logistic Regression, Random Forest, Gradient Boosting, LightGBM, A/B testing, hypothesis testing, feature engineering Time Series: Prophet, SARIMA/ARIMA, walk-forward cross-validation, anomaly detection AI: Anthropic Claude API, RAG, sentence-transformers (embeddings, semantic retrieval), Streamlit, prompt engineering Analysis: RFM segmentation, cohort retention, churn, demand forecasting, KPI dashboards CRM: Salesforce Agentforce (Trailhead-trained) Tools: Git, Jupyter, VS Code
AI Resume-JD Matcher - live web app Scores resume-JD fit using Claude API and semantic embeddings. Returns match score, missing keywords, suggested bullet rewrites, and gap analysis. Stack: Streamlit, Claude API, sentence-transformers, pdfplumber, RAG Live demo · Repo
Multi-Entity Retail Analytics Consolidated sales, margin, and inventory-shrink analysis across a 3-entity, 12-location retail group, unifying two POS systems (Square + PayAnywhere) into one reporting layer. Per-entity and network roll-ups, category margin, and an audit-lens shrink model that separates true loss from receiving variance and count corrections. Stack: Python, pandas, Plotly, Streamlit, Jupyter Synthetic dataset modeled on a real multi-entity retail back-office workflow.
Demand Forecasting + Anomaly Detection Time-series forecasting on Rossmann Store Sales (1M rows, 1,115 stores). Three models compared with walk-forward cross-validation: LightGBM reached 8.20% MAPE, roughly 2x better than Prophet (17.16%) and SARIMA (17.95%). Anomaly detection on residuals. Stack: Prophet, SARIMA, LightGBM, statsmodels, Plotly
Live Tableau Dashboards
- Credit Risk Portfolio - 5K loans, $91.7M, Grade F 30x default rate
- Payment Fraud Detection - 50K transactions, ATM 2x POS fraud rate
- Insurance Claims Analysis - 15K claims, $78.4M, Mental Health 22% denial rate
- 2022-present - Data Analyst, Damir Bary Inc.
- 2020-2021 - Freelance Data Analyst (independent clients)
- 2020 - Contract Data Analyst, Volkovgeology JSC (Kazatomprom subsidiary)
- 2015-2019 - Internal Auditor, NAC Kazatomprom (LSE-listed uranium producer)
- Santa Monica College, transferring to UC Berkeley CDSS (Data Science B.A., target Fall 2028)
- Salesforce Agentblazer Champion 2026 - Agentforce Builder, AI agents, prompt engineering, LLM grounding
- Kaggle - Pandas, Advanced SQL (2026)
- Google Data Analytics - Foundations, Ask Questions (2026)
- Salesforce Trailhead - 25+ modules including Agentforce, Einstein Trust Layer, Generative AI, NLP, Flow Builder, Prompt Builder (Superbadge)
All projects
| Project | Tools | Description |
|---|---|---|
| AI Resume-JD Matcher | Streamlit, Claude API, sentence-transformers | Live web app: resume-JD semantic match scoring with bullet rewrite suggestions |
| Multi-Entity Retail Analytics | Python, pandas, Plotly, Streamlit | 3-entity / 12-location retail roll-up across two POS systems, margin + audit-lens shrink analysis (synthetic data) |
| Demand Forecasting + Anomaly Detection | Prophet, SARIMA, LightGBM, Plotly | Rossmann 1M rows, walk-forward CV, LightGBM 8.20% MAPE |
| Credit Default Prediction | Python, scikit-learn | Loan default model, LR vs RF vs GB with feature engineering and threshold tuning |
| A/B Testing Case Study | Python, SciPy, statsmodels | Checkout conversion A/B test with power analysis and significance testing |
| SQL Case Studies | PostgreSQL | 5 advanced SQL patterns: cohort retention, Top N per group, running totals, gap-and-island, LTV |
| Credit Risk Portfolio Analysis | Python, SQL, Tableau | Default rate segmentation by grade, income, vintage cohort |
| Payment Fraud Detection | Python, SQL, Tableau | Fraud pattern analysis by channel, merchant, time-of-day, geography |
| Insurance Claims Analysis | Python, SQL, Tableau | Claims cost and denial rate by plan, provider, denial reason |
| E-commerce Customer Segmentation | Python, SQL | RFM, churn, and cohort retention for a French fashion retailer |
| Drilling OPEX Analysis | Python, SQL, Power BI | Operational cost analysis for uranium drilling across 12 deposits in Kazakhstan |
| Supply Chain Delay Analysis | Python, SQL, Power BI | Shipment delay root cause across ports in Southeast and East Asia |
| Marketing Campaign Dashboard | SQL, Python, Power BI | ROAS, CTR, conversion rate across channels and campaign types |
| Retail Sales Analysis | SQL, Python | Revenue and customer behavior analysis of POS transactions |
| MedTransport BI Analytics | Python, pandas, Jupyter | Synthetic NEMT company: KPIs, cohort retention, channel performance, claims, unit economics |
| Social Media Engagement Analysis | SQL, Python | 10K posts across 5 platforms: engagement drivers, content type, campaign effectiveness (synthetic) |