● Rupert: A conversational assistant to find restaurants in San Francisco. RAG design that combines blogs, Yelp, and search results. Uses Langchain and NodeJS. Frontend hosted on Vercel, backend is AWS Lambda functions, vector database is Singlestore.● Guess the Card: Evaluation of model reasoning using agents. Programmed a two player game (each an LLM agent) as well as automated labeling. Primary tech is Langchain and Python. LLMs used: GPT 3.5,4; Llama2-7b-Chat, Llama2-70b-Chat. Fine tuning to boost performance.● NBA Analysis Assistant: Created an agent to assist with analysis of NBA teams and players. Agent is capable of navigating and scraping up-to-date data and also provide analytical insights.

● Architect data pipeline to provide a unified view of data across SaaS tools for GTM teams. Maintain transformation, and publishing tools; working with dbt, Snowflake, Looker, Pocus, Hightouch, and Stitchdata.● Technical implementation of Pocus, a SaaS tool from GTM teams. Combines product and CRM data into a single view, the “first tool” used by sales reps.● Built alerting system for dbt, routing table status to owner using Airflow and Slack. We had >2k tables and >50 contributors on the dbt repo. Alerts were structured with actionable information and routed to the relevant owners.● Contribute to dbt platform tooling used by employees across the company. This includes CI/CD, and local development tooling.● Refactored Account models for GTM teams. Work with data analysts, marketing, and sales teams to standardize data flow for account segmentation and assignment.

● Lead development of data platform using Fivetran, Segment, Snowflake, dbt, and Looker.● Work with product, operations, and marketing teams to identify, define, and implement changes to data infrastructure.● Build and maintain integrations with SaaS providers.● Monitor and maintain backend infrastructure for performance and reliability; focused on MongoDB, GraphQL, AWS environments (ElasticBeanstalk, Elastic Container Service, Lambda)

● Responsible for full data stack, connecting SaaS services to proprietary services within AWS.● Built and managed marketing reporting infrastructure; provide ROAS reporting for ad spend optimization.● Managed vendor relationships; researching and integrating solutions for data pipeline.

● Maintained and extended core ETL processes. For various DAGs, achieved 4x runtime through parallelization configuration, ~20% reduction in duration by using task weights, and achieved ~100% stability with caching.● Extended monitoring and alerting infrastructure, providing rich context for Airflow incidents.● Built system dashboards for AWS assets and Airflow DAGs using DataDog and Periscope providing richer information that was easier to consume over service defaults.● Introduced operational best practices, such as On-Call Playbook and System Performance Reviews, to encourage collaboration and cross-training within the team and proactively identify system issues.● Administered Postgres data warehouse, including performance tuning and error monitoring.● Launched initial integration of Looker to provide self-service analysis for non-SQL users. ● Collaborated with and coached junior team members on SQL, Python, and data modeling best practices.

● Work with product and customer support teams to define new features and data model for ad-hoc reporting.● Architected and built streaming ETL using Spark on YARN and Kudu providing transformed and enriched data within a 1 second SLA.● Kept unit and integration test coverage of new pipeline >80%. Integrated Docker for testing dependent services.● Deployed and administered Apache Superset, providing interactive analytics capability across all departments and technical skill levels.● Designed OLAP schemas to facilitate drill down to individual events; leveraged Apache Kudu and PostgreSQL as primary data stores.● Developed and maintained nightly batch scripts populating nightly roll-up tables; scheduled with Apache Oozie.● Defined and implemented customer reports served via Golang web service.● Extended Storm streaming job to perform full Kafka replay from S3 backups.● Scrum lead and organizer of Parsable's Engineering Book Club.

● Responsible for entire lifecycle of product; ideation, implementation, and iteration of features.● Engineered social platform utilizing iOS, Swift and self-hosted Parse where users organize, discover and share local restaurants. Primary responsibilities were back-end, architecture, and product direction.● Built distributed web scraper in Python, Scrapy, and MongoDB running 25 slave machines processing 600+ successful HTTP requests per minute at peak with headroom. Managed fleet with Ansible.● Constructed NLP recommendation engine from 2M restaurant profiles and 30M reviews to quantify and predict user taste profiles utilizing Python and NLTK.● Prototyped and architected search engine using Python, NLTK, and MongoDB to provide natural language recommendations based on taste profiles; maintained average 2mS query time.

● Operated short term rental in 2 different locations within San Francisco.● Responsible for guest interaction and satisfaction. First point of contact for guests and ensured smooth check in and check out process.● Optimize placement within AirBnB platform, including search optimizations and pricing.● Negotiated and executed lease and revenue share agreement with property owner.● Managed vendor relationships; including maintenance, legal, and operational services.● Cleaned. There was a lot of cleaning. So much cleaning.

● Prevented 50% drop in revenue through developing LTV (lifetime value) model using R to balance organic and paid marketing channels against download and in-app revenue.● Developed ETL (extract transform load) from AWS Redshift to PostgreSQL for publisher interactive dashboard optimizing for sub-second page response.● Migrated game studio data platform from 3rd party analytics system to AWS Redshift, greatly reducing query times by 10X while providing efficient multi-dimensional data analysis.● Prototyped multi-PB real-time data analytics streaming system using Kafka and Spark to aggregate information from millions of devices, achieving throughput of 1 GB of messages per second.

● Consolidated 40+ Tableau workbooks into single entity by writing ETL in SQL to run > 25 views from 3 data sources and standardizing metric definitions for PMs and Game Designers to facilitate better decision making. ● Increased event gross revenue by 20% through identifying improperly recorded metrics and ensuring accurate delivery to Game Designers.● Standardized core schema for consistent reporting across all titles, decreasing implementation time by 50%.

I was responsible for the entire lifespan of data for Recharge Studios titles, from client to server to reporting to analysis. Used data to decreas acquisition costs, increas retention, and find compelling monetization solutions. Used data to ensure users were provided with the best experience possible.

Sit for Something used movie theaters to raise money for educational organizations. Good movies, good causes. We were a young company hoping to make the world a better place, one feature film at a time.

I worked with cross functional teams to research business and data structures and translate those specifications into functional systems. My responsibilities focused on developing and maintaining reports and dashboards using SQL-based technologies. By working closely with business leaders and end users, I helped define and implement solutions tailored to specific business needs.