Senior MLOps & AI Engineer

5/2025 – 3/2026

- Led end-to-end delivery of the ML and AI applications; drove cross-team collaboration across front-end, back-end, DevOps, and ML teams and ensured successful integration of these applications into the QuantistryLab simulation platform
- Designed and implemented an agentic compound search system on Kubernetes that uses an open-weights LLM, the Pydantic AI agent framework, the Model Context Protocol (MCP), and a Postgres database
- Reduced infrared spectra prediction time by 84%, from one hour to 10 minutes, by replacing a quantum mechanical model with a graph neural network
- Established model lifecycle management by implementing model registry and versioning practices
- Introduced infrastructure for serving LLMs on Kubernetes
- Designed CI/CD pipelines for monorepositories to deploy ML applications across staging and production environments

MLOps, Docker, Machine Learning, Kubernetes, Data Science

8/2023 – 4/2025

- Led ML engineering practices for 2 squads, designing ML pipelines, deploying them to production, implementing monitoring and model lifecycle management on Databricks
- Improved codebase quality and reduced PySpark runtime by up to 50% by refactoring overly complex code, modularizing data pipelines, eliminating non-operational logic, and enforcing automated quality checks via CI/CD
- Streamlined the development workflow by implementing the gitflow branching strategy and a monthly release process

Databricks, MLOps, Machine Learning, Data Science

4/2020 – 7/2023

- Automated income and expense verification for loan applications by developing ML models on Google Cloud Platform (GCP) with PySpark, Airflow, and Azure CI/CD, enabling up to 70% of loans to be processed without manual review
- Trained classification models with Random Forest and Gradient Boosting on imbalanced transactional data
- Built an interactive exploratory dashboard for analyzing mobile app reviews, powered by a text classification model
- Facilitated the adoption of standardized data pipelines by conducting a company-wide workshop on Kedro-based data engineering practices

Scikit-learn, Data Science