the résumé of adam patrick georgiou

Nov 2023 → Present Principal Software Engineer @ Runway.com

Early engineering hire (#12) at this FP&A platform; helped grow from seed (a16z) to Series A (Initialized, Garry Tan), 50+ employees, and millions in ARR. Led major initiatives across the stack in partnership with the CTO and CEO.

Nov 2014 → Nov 2023 Senior Software Engineer @ Google

Built large-scale collaboration and web platform systems across Google Drive Web and DoubleClick Ads Verification. Left voluntarily in good standing to pursue startup work.

2010 → 2014 Software Engineer @ FactSet Research Systems Inc.

Real-time search and alerting over market data on a C++, Linux, and Elasticsearch stack — one of Elasticsearch's earliest production users.

2023 Machine Learning Experiments

A series of experiments building the canonical MNIST Neural Network, some toy RNNs and CNNs. Implemented from scratch, in C++, to test my understanding of the math; and in PyTorch, to learn the basics of its API. More detail at github.com/georgi0u/ml_experiments.

2022 → Present Deck of Cardio (.com)

Flutter & Firebase based mobile app for working out with a deck of cards.

Financial Modeling Agent — "ARI" @ Runway.com

• Led a team of three to design and build an LLM-powered assistant in Go and React that lets users query, explain, and manipulate highly dimensional financial models through conversational workflows.
• Owned the execution loop end to end: session management, planning and orchestration, tool invocation, permissioning, context assembly, structured outputs, and the React UI.
• Improved reliability, responsiveness, and cost via prompt caching, context compaction, model selection, and iterative prompt development with Braintrust.

"Ambient" AI @ Runway.com

• Designed, patented, and built an asynchronous LLM processing pipeline that precomputed contextual explanations over customer financial data and surfaced them proactively in-product.

Core Data Platform @ Runway.com

• Architected and led a team of five to implement a PostgreSQL-backed, versioned datastore with a custom Go ORM/SDK, integrity constraints, and optimized throughput; tailored to highly dimensional financial data.
• Executed a live, incremental, feature-flagged migration from a bespoke datastore at ~100M-row scale, with zero downtime and zero data loss.
• Made the case for the redesign to leadership and authored the canonical engineering documentation for the new system.

Integrations @ Runway.com

• Designed, built, and led a team of three on Runway's ETL pipeline, ingesting hundreds of financial data sources (accounting, payroll, banking, CRM, ad platforms) via Fivetran, Merge.dev, Workato, and direct connections.
• Hundreds of GB and hundreds of thousands of API calls per day through a bespoke orchestration system handling scheduling, retries, schema evolution, and monitoring; backed by Go services, Snowflake, and PostgreSQL.

Production Reliability @ Runway.com

• Owned reliability for Runway's GCP deployment — Kubernetes, Redis, Cloud SQL — including observability, on-call rotation, and lead response on sev-1 incidents.
• Improved debuggability via custom Sentry, Datadog, and PagerDuty instrumentation, and reduced deployment risk with a blue/green rollout strategy on Kubernetes.

Collaboration & Productivity Platform @ Google

• Tech lead for Google Drive Web's offline infrastructure, using Service Workers and local storage to make Drive and Docs usable without a network for over 100 million daily-active users.
• Built Apps Script APIs allowing first- and third-party applications to extend the Drive web frontend; used by over a million developers.
• Migrated Drive Web to server-side rendering (~30% page load improvement) and managed the company-wide migration of Drive and Docs to a new identity management system across dozens of services.
• Led a two-engineer rewrite of the DoubleClick Verification application from GWT/AngularJS to AngularDart.

Search & Information Retrieval @ FactSet

• Built real-time search and alerting over market data — thousands of sources, petabytes of data, hundreds of QPS — on a C++, Linux, and Elasticsearch stack.
• Implemented company-wide Python tooling from scratch: BuildBot integration, setuptools, pylint, an internal PyPI server, and pip.

Rensselaer Polytechnic Institute

Dual B.S. Degree in Computer Science & Marketing with a GPA: 3.9.

• Languages: Golang, TypeScript, JavaScript, Dart, Python, Java, C++
• Frameworks: React, Angular, Closure, Flutter, NumPy, Pandas, PyTorch, Django
• AI: OpenAI/Anthropic APIs, Braintrust
• Tech: GCP, PostgreSQL, Kubernetes, Firebase, Snowflake, Vercel, Sentry, Datadog
• Integrations: Fivetran, Workato, Merge.dev