Data Scientist with 5+ years of experience driving product and business decisions through experimentation, causal inference, and statistical modeling at scale.
- 🏢 Currently Data Scientist @ Walmart Connect — building Brand WAMM models, NLP pipelines & causal inference frameworks
- 🎓 M.S. Computer Science — California State University, Long Beach (2021–2023)
- 🤖 Deep expertise in NLP, Transformer models, MLOps and multi-cloud AI deployments
- 📊 Track record: 60% ↓ tagging effort · 40% ↓ runtime · 25% ↑ marketing effectiveness
- 🏆 Active Kaggle competitor — sharing notebooks, datasets & competition solutions
- 📍 Farragut, Tennessee · Open to Data Scientist & ML Engineer roles
I'm actively competing in ML challenges and sharing reproducible work with the Kaggle community:
- 🏁 Competitions — End-to-end ML/DL solutions with leaderboard results
- 📓 Notebooks — EDA walkthroughs, feature engineering guides & model experiments
- 📦 Datasets — Curated open datasets published for the community
- 💬 Discussions — Tips, insights & collaboration with fellow Kagglers
⭐ Visit kaggle.com/pathik1511 for my latest notebooks and competition results.
| Project | Category | What it does | Stack |
|---|---|---|---|
| 🏆 LLM Prompt Recovery | NLP · LLM | Recovers hidden rewrite prompts from Gemma input/output pairs via perplexity-based candidate ranking — Kaggle competition | Python · Transformers · open-llama-7b |
| 🏆 CommonLit — Evaluate Student Summaries | NLP · Transformers | Two-target transformer regressor scoring student summaries on content & wording, with prompt-grouped CV — Kaggle competition | Python · DeBERTa · HuggingFace |
| ☕ Coffee Sales Forecasting | ML · Forecasting | Sales prediction & inventory optimisation with classical ML models | Python · Scikit-learn · Pandas |
🏆 = Kaggle competition solution · More at kaggle.com/pathik1511
60% reduction in manual tagging effort → spaCy & Hugging Face pipelines (Walmart Connect)
40% runtime reduction → Production WAMM framework in Python/SQL
25% boost in marketing effectiveness → Sentiment analysis, Naïve Bayes (Syntrons)
15% creative effectiveness improvement → NLP on customer reviews (Walmart Connect)
35% fraud detection improvement → Deep learning fraud detection (Syntrons)
50% data processing speed increase → PySpark big data optimisation


