I work where technical decisions become business consequences.

Replatformed a high-cost vendor-built system, reduced code surface by 60%, and led the build-vs-buy path for its replacement.

Inherited a vendor-built platform with high cost, low velocity, and fragmented architecture. Consolidated microservices into a core system, replatformed infrastructure with IaC, and established engineering standards across the team. Audited the vendor landscape and proved the in-house path via a three-week prototype, leading to a full rebuild and migration. Closed a seven-figure cost center and brought the system back under direct control.

Architected the platform layer for KPMG’s newly-formed AI practice — its fourth pillar alongside Tax, Audit, and Advisory.

AI work across KPMG was fragmented by function, client, and geography, with no shared visibility or reuse. Built the system of record for AI tools, agents, and solutions across global member firms, including onboarding, governance, and discovery. Enabled partners to reuse proven capabilities in client work and RFPs, reducing duplication and raising the baseline of delivery.

Took a comp-equity and transparency platform from zero to one alongside founder Roger Lee.

Architected and built the product end-to-end with the founder: ingestion across Carta and industry payband sources, anomaly detection across pay bands, and configurable comp-review workflows with multi-stage approval hierarchies. Helped hire the founding product, engineering, and design team and onboarded the incoming CTO. Early customers included HopSkipDrive and LaunchDarkly.

Built the control plane behind NextEra 360; 10,000+ devices enrolled, 500+ MWh deferred in one summer.

Orchestrates tens of thousands of distributed devices — home batteries, EVs, smart thermostats — against real-time market pricing and customer-set availability. Now powers customer deployments including Gexa Energy: 10,000+ devices enrolled, 500+ MWh of peak load deferred in a single summer.

Productized the CRISPR pipeline so edits batch by reagent compatibility — 10× throughput.

CRISPR gene-editing service was capacity-bound: scientists worked cell edits one petri dish at a time. Productized the lab’s core pipeline so like-type edits could run as batches grouped by reagent compatibility and urgency, rather than one customer request at a time. Throughput up 10×; revenue up an order of magnitude.

Forked Kibana into investigator-grade tooling — turned forensic storytelling into a competitive edge.

Forensics practice had all the data they needed in a large Elasticsearch cluster but no way to reconstruct cohesive attack narratives. Built origin-of-event tracing using internal heuristics, cross-instance real-time search, and a bespoke data-visualization layer for explaining attack timelines — exportable directly into client deliverables. Cut investigator workload by an estimated two orders of magnitude.

Built a firm-wide automation platform replacing millions of manual hours and standardizing execution across tens of thousands of practitioners.

Architected a platform that unified the core automation stack (Alteryx, UiPath, Blue Prism, Python, Excel/VBA), letting practitioners upload inputs, run automations at the click of a button, and receive verified outputs. Jobs executed in isolated sandboxes that scaled with demand. KPMG rolled it global across member firms.