Data Scientist | NLP & LLM Engineer | NYU M.S. Data Science '26

📍 New York, NY  ·  📧 [email protected]  ·  💼 linkedin.com/in/deepali-bk  ·  🌐 Portfolio

I'm a Data Science graduate student at New York University (GPA: 3.6) with a background in software engineering and machine learning. My work sits at the intersection of NLP, LLMs, and production ML systems — from processing millions of unstructured documents to evaluating emotional intelligence in large language models.

I've built classifiers, information extraction pipelines, and multilingual NLP systems across healthcare, enterprise, and research domains. I care about models that work in the real world, not just on benchmarks.

• 🔬 Graduate Research Assistant @ NYU Rory Meyers College of Nursing
• 🏆 Violet Internship & Research Award 2025
• 🌐 Women in Data Science Ambassador 2026

Languages & Tools Python R SQL C++ Git

Machine Learning & Deep Learning Scikit-learn PyTorch TensorFlow XGBoost Flair FastText Computer Vision

NLP & Generative AI HuggingFace Transformers LangChain SpaCy BERT RoBERTa GPT-4 RAG Systems Sentiment Analysis Topic Modeling Named Entity Recognition Few-shot Prompting

Data & Visualization Pandas NumPy Matplotlib Seaborn Tableau Spark Hadoop (HDFS)

Statistics Statistical Modeling Bayesian Methods A/B Testing Hypothesis Testing

Production-grade ML pipeline combining BeautifulSoup + XGBoost with rule-based heuristics for HTML quality detection.

• Achieved 0.98 precision / 0.920 F1 score
• Built a GPT-4.1-mini few-shot labeling pipeline to expand an imbalanced seed dataset
• Eliminated manual QA bottlenecks entirely

Python XGBoost BeautifulSoup GPT-4 Few-shot Learning ML Pipelines

Benchmarked emotional intelligence capabilities of Gemma, Qwen, and Llama using zero-shot and few-shot prompt engineering.

• Demonstrated above-chance performance across all three models
• Leveraged LangChain and HuggingFace for evaluation orchestration

LangChain HuggingFace Prompt Engineering Zero-shot Few-shot LLM Evaluation

• Statistical modeling on survey data from 90+ countries for COVID-19 healthcare pattern recognition
• Built interactive Tableau dashboard for the organization's public-facing website
• BERT-based multi-class classifier achieving 87% accuracy on 5,000+ survey responses on violence against medical professionals

• Analyzed 21,000+ survey responses across 40+ languages using multilingual RoBERTa
• Applied sentiment analysis and topic modeling on PPE usage data during COVID-19

• Led NER implementation as SME for a Fortune 500 telecom client — extracted entities from 2M+ unstructured documents using SpaCy
• Reduced manual review time by 71% through automated document processing
• Built classification system with Flair + FastText to categorize 10,000+ reviews for C-suite strategic planning

Relevant coursework: Deep Learning · Machine Learning · Natural Language Understanding · Reinforcement Learning · Big Data · AI Applications in Business (NYU Stern)

• 🥇 Violet Internship & Research Award 2025 — Competitive funding for excellence in research and internship performance
• 🌟 Women in Data Science Ambassador 2026 — Selected to represent and promote WiDS initiatives

Always open to research collaborations and data science opportunities. Feel free to reach out!