Springbig, Inc.
May 2022 - Present- Business Analyst
- Software Engineer
- Team Lead
- Head of Engineering
Advanced from Business Analyst through Software Engineer and Platform Team Lead to Head of Engineering for a white-labeled enterprise messaging platform, by aggressively triaging neglected technical debt, directing deployments and change management, leading rapid-response remediation, and serving as the standing liaison between engineering, finance, product, customer success, and enterprise clients.
- Resolved an existential domain-reputation crisis where major security vendors flagged merchant domains as suspicious; traced the root cause to legacy Cloudflare Workers regex failures under bot crawling and shipped a surgical fix in under 24 hours to fully restore platform trust.
- Architected a high-performance custom audience segmentation engine using PostgreSQL JSONB containment queries and inverted indexes, replacing expensive join-table chains to drastically improve evaluation speed, predictability, and maintainability.
- Rescued platform stability across microservices following a major end-of-life Ruby/Rails upgrade by rapidly identifying deep dependency failures and deploying targeted patches across teams in a single night.
- Enforced cross-repository CI/CD compliance by engineering a declarative configuration script that actively detects and autonomously corrects pipeline and merge-check drift across dozens of GitHub and Bitbucket repositories.
- Cut merchant dashboard first-load download by 94% (5.3 MB → ~300 KB) in a one-week frontend overhaul: retired sunset product code, split the monolithic bundle by design system, vectorized raster assets, and moved ~250 routes to on-demand chunks — halving total JavaScript shipped and tripling production build speed.
- Built out messaging compliance and sender infrastructure: secured Campaign Service Provider registration, replaced the manual 10DLC workflow, introduced 1:1 sender-identity minting that drove domain churn to near zero, and moved email authentication to enforced policy.
- Led incident response and remediation of silent platform failures — a full native-app outage affecting a top-five account, a blocking webhook path costing four figures monthly, and a POS integration broken since release — while formalizing the release process and cross-training engineers to remove a key-person dependency.
- Serve as the primary engineering liaison for executive leadership, finance, vendors, and clients during critical escalations, seamlessly translating business risk into technical execution.
- Averted major legal and compliance risk within an acquired Laravel/PHP product when SMS opt-out enforcement failed; diagnosed a Year 2038 timestamp overflow in a 32-bit MariaDB column and deployed a critical, fully compliant hotfix.
- Rescued an enterprise sales commitment by completely redesigning the platform's legacy rewards tiering model, enabling seeded tier initialization and frictionless onboarding for a major client.
- Exposed the redesigned rewards tiering architecture via external APIs and webhook flows backed by materialized views and staging tables, providing clients real-time access to spend totals without expensive runtime computation.
Expand for technical contextHide technical context
Domain reputation crisis
When major security vendors and browsers began flagging merchant-facing domains as suspicious, the issue escalated from a nuisance into an existential trust problem for a white-labeled platform serving regulated businesses. The root cause was eventually traced to brittle regular expressions in legacy Cloudflare Workers that threw exceptions under bot crawling, making legitimate domains appear unsafe to automated scanners. A surgical fix resolved the issue in under a day and led to a more durable framework for platform trust signals going forward.
Segmentation refactor
The custom audience builder had accreted into inefficient, difficult-to-reason-about query chains. Analysis of real merchant behavior showed that most “custom” audiences were recurring combinations of a small set of reusable base conditions, not truly unique queries. The redesign shifted the service toward precomputed base segments and reusable named groupings, with PostgreSQL JSONB containment queries and inverted indexing replacing heavy join-table chains for common evaluation paths.
Frontend delivery overhaul
Every merchant visit previously began with a single 5.3 MB compressed bundle (13.5 MB parsed) re-downloaded after every release. A one-week initiative retired a sunset feature surface and its dependencies, split the bundle so UX2 and UX3 merchants each download only their own design system, replaced four multi-hundred-kilobyte “vector” images that were actually embedded photographs, and converted roughly 250 routes into independently cached on-demand chunks. First-load dropped from 5,329 KB to about 303 KB, total shipped JavaScript fell 47%, production builds went from over three minutes to about one, and error monitoring was repaired for half the merchant base. Full write-up and segmentation maps →
Sender infrastructure and registry standing
Following an extended internal effort, the company was provisionally accepted as a Campaign Service Provider with The Campaign Registry, enabling direct migration of client brand and campaign registrations between connectivity partners, direct access to aggregator data, and additional registry API endpoints. In parallel, the fully manual, error-prone 10DLC registration workflow on a business-critical path was replaced with an end-to-end redesign. Reference: the 10DLC regime, reconstructed from primary sources → A sender-identity minting process established a strict 1:1 relationship between sender and domain, replacing a policy of continuous domain rotation; short-domain churn fell to near zero, recurring registration spend on rotated domains was eliminated, and client sending domains were no longer exposed to automated discovery.
Email authentication and domain trust posture
SPF and DMARC records for company property domains were identified as misaligned and corrected in coordination with IT and DevOps, fixing authentication failures across platform email and moving sender identity from published policy to enforced policy. Case study: interactive DMARC policy simulator → For client-facing domains, a trust-signal standard was established: elimination of request fall-through to the primary platform, search-engine domain verification, a stricter x-robots and robots.txt policy, and a required security configuration checklist.
Native application outage
A full outage of the native applications affected enterprise merchants, including a top-five account. A three-person investigation ran through the weekend and was escalated to a broader team when the cause remained unidentified. The root cause was isolated to a pre-authentication initialization file served from an S3 bucket absent from any current inventory, retired after more than a year without updates. A patch was deployed within thirty minutes of diagnosis.
Silent integration failures
A sustained async-worker compute spike following an upstream e-commerce platform change was traced to a blocking synchronous call in the webhook path, with roughly 25,000 members affected and growing and DevOps estimating four figures per month in additional AWS spend; it was remediated by closing a redundant dual path and replacing the blocking call with an event-driven post. Separately, a point-of-sale integration was found never to have functioned end to end since release about ten months earlier, leaving five enterprise merchants with visit records lacking customer detail; the defect was corrected, the integration’s development lifecycle reconstructed, and affected merchants restored through a configuration update coordinated with Client Services.
Release process and operational hygiene
Ticketing, workflow, and deployment processes were documented and formalized end to end, and two engineers were trained to execute deployments independently, removing a key-person dependency from the release path. An expired SAML signing certificate on the identity provider was also identified ahead of remaining service-provider rollovers, preserving federated authentication across all managed services.
Repo governance and upgrade recovery
A separate stream of platform work focused on reducing drift and operational fragility across the engineering estate. CI/CD and merge policies were re-aligned across GitHub and Bitbucket repositories through declarative, periodically enforced configuration, and a bad Ruby/Rails version bump was stabilized by recognizing a common dependency root cause behind multiple scattered breakage reports and patching the gaps in a single night.
Year 2038 opt-out failure
In an acquired Laravel/PHP product with little in-house expertise, SMS opt-out enforcement silently failed and created immediate legal exposure. Investigation alongside the technical founder traced the issue to a real-world Year 2038 overflow in a 32-bit MariaDB column used to represent a far-future “next eligible send” date, and a pragmatic fix restored compliant behavior until the product could be sunset.
Rewards tiering redesign
A major enterprise opportunity exposed gaps between promised functionality and what the legacy rewards tiering model could actually support. The tiering capability was redesigned to allow seeded tier initialization, reduce reliance on brittle legacy behavior, and expose real-time spend and visit data through external APIs and webhook flows backed by materialized views and staging tables.