Data Analyst Research Assistant — Applied AI / MLOps
May 2025 – Present
Applied AI & ML Infrastructure
- Orchestrated ML training workflows across Vertex AI and Snowflake; introduced Git-based CI/CD automation to standardize runs and reduce training time by 2 hours per cycle.
- Operationalized reproducible experimentation by versioning data extracts, feature definitions, and training artifacts to improve traceability and reviewability in research iterations.
- Built feature engineering pipelines (Python/SQL) and collaborated with stakeholders to translate research requirements into production-ready ML features.
Analytics & Cost Optimization
- Developed exploratory analysis in Tableau to surface signal quality issues, guide feature prioritization, and accelerate model iteration.
- Optimized BigQuery workloads through partitioning and clustering strategies, driving a 38% reduction in query costs while preserving analytical fidelity.
Applied AI: LLM-Enabled Analytics (Research)
- Prototyped retrieval-augmented analysis patterns (embeddings + semantic retrieval) to enable natural-language exploration of datasets, documentation, and research outputs.
- Established lightweight evaluation checks (e.g., relevance and consistency) to keep model-assisted outputs auditable and aligned with source data.
Faster ML cycles
2 hours saved per training cycle via CI/CD + standardization
Lower analytics spend
38% BigQuery cost reduction through physical design optimization
Vertex AI
Snowflake
BigQuery
Python
SQL
MLOps
RAG
Embeddings
LangChain
Tableau
CI/CD