Brian Kutsop
Staff Software Engineer
Staff software engineer with over 11 years building large-scale, fault-tolerant distributed systems and shipping production AI. I'm the lead engineer for Intuit's most critical contact-center platform — 8.4M customers, 61M voice contacts, and 1.35B API requests a year, with zero P0–P2 incidents for over a year. I design systems for reliability, engineer effectively under ambiguity, build hands-on with LLMs and agentic tooling, and bridge partner companies, legal, architecture, and engineering teams.
Technical Skills
Languages: Java, TypeScript / JavaScript, Python, C# / .NET, SQL, Swift
Distributed systems & cloud: AWS (multi-region active-active & disaster recovery, DynamoDB global tables, Lambda, CDK, S3), high-throughput APIs, event streaming, fault tolerance
Applied AI / LLMs: structured-output prompting, retrieval/doc systems for AI agents, Claude Code & Cursor agentic development
Frontend & data: React, GraphQL, observability & dashboards (Splunk)
Experience
Staff Software Engineer — Intuit
San Francisco, CA · 2021 – Present
Promoted Software Engineer (2021) → Senior Software Engineer (2022) → Staff Software Engineer (2026). Lead engineer for the VEP contact-center platform powering all voice, chat, and video communications for Customer Care, Sales, and HR across TurboTax, QuickBooks, Credit Karma, and Mailchimp.
- Owns the platform that can't go down. Lead engineer and primary technical point of contact for a system serving 8.4M customers, 61M voice contacts, and 1.35B API requests per year; sustained zero P0–P2 incidents for 1+ years while leading a team of 11 engineers.
- Re-architected the contact center for multi-region resilience. Drove an active-active / disaster-recovery redesign across 25 services, cutting outage recovery from ~3 hours to ~5 minutes (~97%); ran monthly 24-hour live region-failover gamedays with zero customer impact and partnered with AWS to shape Amazon Connect's Global Resiliency features.
- Built a real-time conversational analytics pipeline (Resolution AI). Pioneered a streaming pipeline that analyzes live conversations to attack the ~20% of contacts left unresolved (a targeted $37–66M in annual cost leakage); included on proposed patent (IDF-14204).
- Used structured-output LLM generation to reclassify contact requests for smarter transfer routing, and built dashboards with AI-generated SQL search and call summaries that opened production data to non-technical users.
- Enabled a $250M revenue target. Built tag-based provisioning to claim and manage 6,000+ business phone numbers at scale, supporting expansion of Intuit's expert and storefront network.
- Created leverage for the whole org. Built an AI-readable documentation framework spanning 31 services so engineers and AI agents can design cross-team architecture in days instead of weeks.
- Chose the simplest safe solution. Deleted 316M call recordings (~2 PB) to meet a compliance deadline via a 15-minute configuration change — zero lines of code.
- Led compliance and cross-org alignment. Led tag-based access-control design for IRS §7216 tax and HR call-recording data with AWS and outside legal counsel, closing major regulatory exposure; serve as the recurring technical bridge across AWS, legal, principal architects, and engineering.
Software Developer → Team Lead — Epic Systems
Verona, WI · 2015 – 2021
- Founded "Satellite," a pull-based software deployment & configuration system that became Epic's default deployment method. Each endpoint agent downloads installations, verifies them by hash against the expected value, applies configuration updates, and reports results to a central console for administrators — scaling installs from dozens per hour to 200,000+ machines across separate domains and unreliable networks. Designed its security (RSA/AES, JWT) and self-update so the agent never needs a manual re-push.
- Promoted to Team Lead: drove scalability work, ran the urgent-issue "shield" rotation (80% of priority issues resolved within 2 weeks), reviewed designs across 30+ projects, and presented new features to 450+ customer representatives at Epic's annual conference.
Selected Project
TravelApp — personal project; agentic, AI-built trip planner. A cross-platform app (SwiftUI / iOS, .NET Core, Azure; Amadeus, Viator, and Google Places integrations) built largely by orchestrating Claude Code agents — parallel /loop and scheduled background orchestrators working a shared task queue, phone notifications via remote control, and adversarial self-review before merge. A working demonstration of designing, directing, and shipping real software through autonomous AI agents.
Education
Cornell University — M.Eng., Computer Science (2015) · B.S., Computer Science, Dyson business minor (2014)