Senior Data Scientist / Machine Learning Engineer, NLP - 1635

Must Have:

• 4–6+ years of data science or machine learning experience
  
• NLP classification for customer messages or call transcripts
  
• Intent, topic, sentiment, and multi-label classification
  
• Confidence scoring and model evaluation
  
• Text cleaning, deduplication, speaker handling, and PII-safe processing
  
• Trend and anomaly detection
  
• Python, PySpark, SQL, and pandas
  
• Labeled dataset design and annotation workflows
  
• Precision, recall, confusion matrix, and drift monitoring
  

Responsibilities:

• Build and deploy NLP classification models for customer communications.
  
• Develop intent, topic, sentiment, and multi-label taxonomies.
  
• Clean and prepare transcript and message data for modeling.
  
• Handle short-text cases, duplicate records, system messages, and speaker identification.
  
• Build trend and anomaly detection methods using baselines, seasonality, and channel mix.
  
• Design maintainable Python and PySpark data pipelines.
  
• Define sampling strategies and annotation guidelines for labeled datasets.
  
• Support reviewer adjudication and dataset quality validation.
  
• Track model precision, recall, confusion patterns, confidence scores, and drift.
  
• Implement secure processing for customer communications containing sensitive data.
  

Qualifications:

• 4–6+ years of relevant machine learning, NLP, or data science experience.
  
• Proven experience deploying NLP models into production.
  
• Strong experience with classification systems and text analytics.
  
• Advanced Python development and testing skills.
  
• Hands-on experience with PySpark, SQL, pandas, and scalable data pipelines.
  
• Experience creating and validating labeled datasets.
  
• Strong understanding of model evaluation, monitoring, and false-alert reduction.
  
• Experience working with governed or PII-bearing data.
  

Nice to Have:

• Databricks
  
• Unity Catalog
  
• Databricks Workflows
  
• MLflow
  
• Model and data versioning
  
• Retrieval and embedding models
  
• LLM-assisted classification with evaluation and guardrails
  
• Contact-center or customer-support analytics
  
• Property-management or real-estate data experience