Data Scientist/ML Engineer
Job Summary:
We are seeking a Data Scientist/ML Engineer to develop and optimize analytics pipelines for large-scale chat conversation data. The role focuses on conversation analytics, insight generation, semantic clustering, text embeddings, and LLM-assisted analysis. The ideal candidate will build scalable data pipelines, develop clustering strategies, generate actionable insights, evaluate analytical quality, and implement automated testing and validation frameworks. The position also requires adherence to organizational security policies and best practices for handling personally identifiable information (PII) and sensitive data.
Key Responsibilities:
Design and implement analytics pipelines to extract meaningful patterns from large-scale chat conversation data.
Develop facet extraction approaches using large language models (LLMs) to categorize conversations by request type, task performed, and topic discussed.
Build dashboards and reporting artifacts to communicate usage trends, emerging topics, and user behavior to stakeholders.
Identify and quantify shifts in conversation patterns over time to inform product roadmaps and content strategy.
Translate analytical findings into actionable recommendations for platform improvement.
Architect and optimize hierarchical clustering pipelines using density-based algorithms such as HDBSCAN to group conversations by semantic similarity.
Generate and manage text embeddings at scale using embedding models for clustering and similarity tasks.
Design multi-level clustering strategies that produce granular groupings and higher-level category taxonomies.
Evaluate cluster quality using persistence metrics, silhouette analysis, and domain-informed validation.
Experiment with clustering parameters, distance metrics, and dimensionality reduction techniques to improve grouping coherence.
Build and maintain Python-based data pipelines for ingesting, transforming, and analyzing conversation datasets.
Develop automated workflows using cloud-native orchestration and compute services to run analytics at scale on scheduled cadences.
Work with object storage, search engines, and relational databases to store and query analytical outputs.
Implement caching, batching, and incremental processing strategies to efficiently manage large embedding and clustering workloads.
Maintain reproducible analysis environments and version analytical artifacts, including models, cluster outputs, and embeddings.
Design and refine LLM prompts for facet extraction, cluster labeling, and conversation summarization.
Evaluate LLM output quality for analytical tasks and refine prompt strategies to improve accuracy.
Leverage model infrastructure for embedding generation and LLM inference.
Explore emerging techniques in LLM-driven data analysis, topic modeling, and automated insight generation.
Develop evaluation frameworks to measure clustering quality, facet extraction accuracy, and analytical pipeline correctness.
Build automated regression tests to detect drift in clustering outputs or degradation in categorization quality.
Validate analytical results against established baselines and domain expertise.
Document analytical methodologies, assumptions, and limitations.
Adhere to technology policies and organizational security controls.
Implement secure coding practices, particularly when handling PII and sensitive data.
Participate in threat modeling and security discussions for API and infrastructure components.
Apply organizational security standards and best practices.
Required Qualifications:
Experience developing analytics pipelines for large-scale conversation or text datasets.
Proficiency in Python for data engineering, transformation, and analytical pipeline development.
Experience with LLMs for facet extraction, text categorization, summarization, and analytical tasks.
Experience with text embeddings, semantic similarity, and clustering techniques.
Knowledge of density-based clustering algorithms such as HDBSCAN, hierarchical clustering, and dimensionality reduction techniques.
Experience evaluating clustering quality using metrics such as persistence metrics and silhouette analysis, along with domain-informed validation.
Experience building scalable data pipelines and automated workflows using cloud-native orchestration and compute services.
Familiarity with object storage, search engines, relational databases, and techniques for processing large-scale analytical workloads.
Experience designing evaluation frameworks, automated regression tests, and analytical validation processes.
Understanding of secure coding practices, PII protection, threat modeling, and organizational security requirements.