• Training the next generation of AI large language models (LLM) using RLHF

• Improving model quality by using best practices to validate and optimize Python code generated by LLMs

• Analyzing outputs of models and providing ratings and justifications for alignment with user-generated prompts

• Collaborating with cross-functional teams to shape the future direction of models

• Led the architecture of key data quality and monitoring frameworks for internal and external data sources and infrastructure including Spark, Glue, MySQL, Snowflake, Airflow, Tableau and third-party vendors

• Mentored engineers and members of cross-functional teams by teaching technical concepts and offering guidance

• Designed the technical interview process and interviewed candidates for data related engineering roles

• Supported the analytics of Aura’s suite of identity protection and security products that helped transform it into a unicorn company

• Developed robust end-to-end batch and real-time streaming pipelines that processed tens of terabytes of daily data orchestrated by Airflow using Python, SQL, Scala, Spark, Databricks, and Snowflake

• Optimized Docker containers for efficient layers that were orchestrated by Kubernetes and integrated with CircleCI

• Fine tuned SQL queries, Spark jobs and clusters for performance gains and reduction in overall costs

• Implemented robust metadata synchronization and data transfer between Snowflake and Databricks

• Migrated from AWS only to Databricks on AWS, Maven to Gradle, Ansible to Terraform, and Zeppelin to Databricks notebooks and integrated Delta Lake

• Developed tools for tracking table-level data lineage and schema evolution in Python and SQL that was key in streamlining the simplification of pipelines and the data model

• Rebranded to Pango and acquired by Aura

• Used Spark, Impala, SQL and Python to process data from Kafka and supported VPN products analytics

• Converted jobs scheduled by Jenkins, cron and Zeppelin to Airflow for unified orchestration

• Migrated from an on-premise Hadoop ecosystem to AWS (S3, EMR, Glue) to improve scalability and performance

• Implemented seamless restart of Tableau Server using iptables and port routing rules that allowed for continuous access across the company

• At Vollmers Lab: vollmerslab.soe.ucsc.edu

• Conducted library preparation and utilized bioinformatic approaches for antibody sequencing data to characterize the adaptive immune system

• Constructed data visualizations based on novel sequencing datasets using Python and Matplotlib

• Leveraged bioinformatic tools such as BLAT to align and map sequenced DNA fragments and ran scripts on Rocks cluster for distributed processing