Resume
Staff-level cloud, data and platform engineer with 13 years across data warehouse engineering, solution architecture, cloud engineering and full-stack development. I spend most of my time on cloud cost and performance engineering through IaC, event-driven distributed systems in Go, and agentic AI wired into real production pipelines.
Skills
- Languages
- Go, TypeScript, Python, Node.js, SQL, .NET Core, Bash
- Cloud & IaC
- AWS, Azure, AWS CDK, Terraform, CloudFormation, Serverless, Docker, Kubernetes, ECS Fargate
- Data
- Snowflake, dbt, Airflow, ETL/ELT, Aurora, MySQL, PostgreSQL, MongoDB, DynamoDB, Redis, Elasticsearch, Kafka
- AI
- Agentic workflows, Claude, AWS Bedrock, HuggingFace, DeepEval, Promptfoo, LLM evaluation
- Frontend
- Angular, RxJS, React, React Native, micro-frontends, Module Federation
- Practices
- Distributed systems, CI/CD, observability, automated testing, design patterns, technical leadership
Selected impact
$13M/yr
saved by an IaC-driven x86 → arm64 Graviton migration
200M+
messages/day through an event-driven Go platform on Kafka and EKS
10TB+
ingested daily into Snowflake at 99.9% pipeline reliability
400%
faster after rescuing a failing legacy PHP system onto Go microservices with zero downtime
62%
lower AI inference cost on an agentic code-review pipeline
2 days
to stand up a compliant cloud environment, down from 2 weeks, via a Terraform module library used across 50+ environments
20,000+
employers served by a benefits platform whose technical direction I own
30%
fewer production bugs after coaching engineering teams onto conventional commits, linting and test-driven delivery
Experience
Staff Cloud & Platform Engineer · Fortune 500 Enterprise
- arm64 Graviton offered materially better price-performance than the x86 instances our legacy production workloads sat on, but nobody was volunteering to be first: the apps were old, and the build tooling could not produce arm64 images at all. I drove the migration through Terraform and coordinated the build side of it, standing up a custom CodePipeline that produces multi-architecture images with docker buildx and caches aggressively between runs. That second piece mattered more than expected — a one-line change to a Java Spring Boot service previously took over 40 minutes to build, push to Nexus, upload to ECR and clear security scanning; the same change now ships in under 10 minutes. With the path proven and the build pain gone, I guided teams across the org onto the pattern: 36% lower hosting costs, 14% better performance, and $13M/year in total savings.
- Built the agentic workflow that made that migration safe: Claude Opus validating arm64-compatible wheels across six production Lambda Layers with an automated reflection loop to catch deployment regressions, at 60% lower token usage.
- Architected an AI code-review pipeline that ingests requirements from the ticketing system and applies per-language rules and custom markdown context — 62% lower inference cost, review turnaround from minutes to seconds, with a second pass that checks the code against the business intent of the ticket.
- Engineered an automated security validation pipeline that red-teams vulnerability patches in an isolated sandbox before deployment, verifying CVEs are actually eliminated. AWS Bedrock + Claude Opus, CodePipeline, DeepEval and Promptfoo.
- Architected cloud-native ELT pipelines ingesting 10TB+ daily into Snowflake, orchestrated with Python and Airflow at 99.9% reliability.
- Built an event-driven billing pipeline on Step Functions, EventBridge, Aurora and S3 feeding a Snowflake warehouse, taking analyst processing from multi-week cycles to one day.
- Engineered a self-diagnosing debug pipeline that fires targeted SQL diagnostics the moment a pipeline step fails, turning hours of manual investigation into a first-pass root cause that arrives with the alert.
- Own technical direction for a group benefits platform serving 20,000+ employers, defining target-state architecture and service boundaries with architects and tech leads.
Senior Software Engineer · Growth Cybersecurity Consulting Startup
- Over a million security alerts a day were arriving from Microsoft Defender, Qualys and CrowdStrike — three vendors, three payload shapes, three severity schemes — and analysts were triaging them by hand, which meant real findings sat behind noise. I independently architected a distributed event management system in Go that normalizes every vendor feed into one common event shape and runs it through a custom rule engine that classifies each alert and fires the matching response workflow. 1M+ alerts a day now route themselves, and analysts spend their time on the findings the engine surfaces rather than on sorting the queue.
- Moved that ingestion off synchronous Lambda onto containers on ECS Fargate provisioned with CDK — 60% lower cost and 80% higher throughput.
- Built a B2B attack-surface management platform in Angular, RxJS, Go and DynamoDB for enterprise financial clients.
Staff Software Engineer · Large Enterprise Healthcare Company
- Physician referrals moved through a system that could not keep up with the volume the business was putting through it, and rewriting it was constrained by the fact that none of its legacy behavior could be dropped — every rule in there was load-bearing for someone. I spearheaded the architecture of an event-driven replacement in Go, Kafka and MongoDB on EKS, tuning partitioning and consumer scaling as throughput climbed, and extended the API to cover the legacy business requirements rather than stranding them. The platform scaled past 200M messages a day with the legacy contract intact and software quality held through the migration.
- Delivered enterprise FHIR applications on NodeJS, NestJS and Angular handling 200M+ daily requests.
- Led the rescue of a failing legacy PHP application, migrating it to Go microservices with zero downtime and a 400% performance improvement.
- An intranet micro-frontend estate had been built by several teams on their own schedules, and it showed: drifting dependency versions, mismatched ECMAScript targets, duplicated components and bundles heavy enough to hurt load times — independently deployable frontends that could not safely compose. I spearheaded the move to Module Federation on Angular, working the hard parts of it: vetting trusted NPM packages, setting bundle-size budgets, tuning webpack builds and standardizing shared dependency versions and build targets across every remote, while driving adoption of the internal design system so teams stopped rebuilding the same components. Teams could ship independently again without version conflicts, builds got faster, and the estate ran on one set of dependency versions and targets with test-driven development as the default path.
Lead Engineer · High-Growth Sleep-Tech Startup
- Led a team of six building a React Native app that pulls sleep analytics off consumer IoT devices — the company’s play against Apple Beddit — backed by a Go and Aurora metrics platform.
- Overnight sleep tracking only earns trust if the recap is waiting when the user wakes up, but a full night of biometrics is far too much to process while someone stares at a loading spinner. I installed and configured the AWS IoT device fleet capturing sleep cycles and cycle length, heart rate, heart-rate variability, movement and snoring, then built a distributed ingest that uploads the night’s readings to Aurora in batched payloads rather than a single blocking write. Users got an instant sleep recap the moment they woke, while the full dataset finished landing behind it — cutting perceived latency to near zero without shortcutting the pipeline.
- Partner companies and internal teams kept asking for the underlying sleepcore metrics, and every request meant hand-built exports from the platform team. I designed and shipped a public metrics API in Go and Echo, fronted by API Gateway over Aurora RDS, structured with the builder pattern for composable queries and deployed on ECS Fargate. External developers and partner companies could build their own dashboards and schedule their own pipeline jobs directly against it, turning a recurring manual ask into self-service and opening the metrics platform as a product surface.
- A direct-to-consumer brand lives or dies on organic search and checkout speed, and both were working against us: the storefronts were aging AngularJS single-page apps that crawlers struggled to index and that made buyers wait on a blank screen, with commerce logic wound into the same codebase. I rebuilt the e-commerce platform as Go microservices on Lambda with Stripe handling payments, then migrated the storefronts onto modern Angular with Angular Universal server-side rendering, backed by a Contentful CMS whose responses were preloaded server-side. Pages became fully indexable and rendered meaningfully faster on first paint, and commerce shipped on its own cadence instead of riding the storefront release.
Senior Software Engineer · Digital Marketing Agency
- The agency’s legacy PHP platform ran on hand-managed VPS boxes — snowflake servers, no infrastructure-as-code, and configuration drift that put client compliance commitments at risk — and the window to get off them was a single month. A lift-and-shift would have hit that date while carrying every one of those problems into AWS, so I designed a target architecture that brought the legacy stack up to cloud compliance instead of replicating the VPS setup: containerizing the services with Docker, rebuilding configuration and secrets handling as environment-driven rather than baked into the host, and deploying onto Elastic Beanstalk so the platform got managed provisioning, health checks and rolling deploys without a bespoke control plane. The full migration landed inside the one-month window as a genuine cloud migration rather than a rehost — reproducible environments, no hand-configured servers left, and a compliance posture the VPS estate could never have met.
- Refactored the legacy PHP API into Go microservices using the builder pattern, adding structured logging, observability instrumentation, unit test coverage and a unified database access layer that standardized query patterns across services. Scaled the API onto AWS ECS with CloudFormation and Auto Scaling groups, giving it horizontal scaling under load and removing manual deployment overhead.
Earlier
- Enterprise Performance Marketing Company — Software Engineer (2015–2017). Built a keyword opportunity platform end to end in Go, MySQL and React for enterprise search clients, lifting analyst productivity 60%.
- Luxury Real Estate Auction House — Web Developer (2014–2015).
Education
Cornell University — Engineering Leadership, minor in Computer Engineering (2020)