Currently: Data Engineer in training → AI Engineering layer. Targeting Germany. Building in public, zero to market.
I build data products and research-grade analysis pipelines. Currently building Rankwell, a GEO/AI-visibility tracker that shows law firms how ChatGPT, Gemini and Perplexity talk about them versus competitors — daily scans, explainable 0–100 visibility scores, citation-gap recommendations.
| Project | What it shows | Stack |
|---|---|---|
| Quantifying Data Quality | Four-dimensional DQ scoring framework (completeness, consistency, accuracy, timeliness), validated against the UCI Air Quality dataset (9,357 observations); NER confidence scoring at 88.4% precision; K-Means clustering on quality metrics | Python · spaCy · scikit-learn |
| German Tech Job Market Intelligence | Production-scale NLP pipeline over 3,200 job postings: multi-source scraping, TF-IDF + spaCy EntityRuler skill extraction (30,324 skill-posting pairs), Levenshtein deduplication, 4 interpretable role archetypes (Silhouette = 0.61) | Python · spaCy · TF-IDF |
| Rider Segmentation & Growth | 5.5M bike-share trips analyzed in BigQuery SQL — casual vs. member behavior with 3 data-driven marketing recommendations | SQL · BigQuery · Tableau |
| Flight Delay Analysis | 3M U.S. domestic flights (2019–2023): delay drivers, seasonality, carrier comparison | Python · Jupyter |
| Return Rate Analysis | E-commerce return-risk segmentation by product category, price range and customer segment, framed against Germany's €92B e-commerce market | Python · BI |
| ESG–GDP Regression | Whether CO₂ emissions per capita and renewable-energy share predict GDP — hypothesis testing and regression diagnostics | Python · Jupyter |
| Manufacturing Cost Variance | Qlik Sense dashboard for manufacturing cost-variance monitoring | Qlik · Python |
- 9,357 records analyzed for the data-quality framework · 96.7% overall quality score
- 3,200 German tech job postings processed · 156 unique skills extracted · 30,324 skill-posting pairs
- 5.5M bike-share rides examined · 3M flights analyzed
- 88.4% NER precision · 0.61 clustering silhouette score
Data science — statistical modeling, NLP and text analysis, unsupervised learning, feature engineering Analytics — SQL/BigQuery, Tableau, Power BI, business intelligence Engineering — Python, data pipelines, web scraping; TypeScript/Next.js for product work Research — hypothesis testing, quantitative analysis, labour-market research, German market analysis
| Portfolio | kandula.studio |
| linkedin.com/in/nikhilvarmakandula | |
| Kaggle | kaggle.com/nikhilvarmakandula |
| kandulanikhilvarma@gmail.com |
Open to collaboration and research partnerships. Last updated: July 2026.