Selected work

Engineering problems, with the outcomes to show for it.

Client names below are withheld under NDA, as is standard for engagements of this kind — the problems, decisions, and results are real.

Analytics / SaaS

Cutting a SaaS dashboard's data footprint from gigabytes to megabytes

Confidential SaaS client

Result

GB → ~178 MB payload

Web AppBackendPerformance

Problem

The product's core dashboard loaded raw, unaggregated datasets on every request. As customer accounts grew, load times stretched into tens of seconds and infrastructure costs climbed with them.

Challenge

The data volume itself wasn't the real constraint — the query and transport strategy was. Every view re-fetched and re-computed data that changed rarely, with no aggregation layer in between.

Solution

Redesigned the data pipeline around pre-aggregation, incremental queries, and a caching layer, and restructured the API contract so the client only ever received what a given view actually needed.

Result

Dashboard payload dropped from gigabyte-scale to roughly 178 MB per session, with load times improving from tens of seconds to near-instant.

B2B SaaS

Designing a multi-tenant SaaS platform from the ground up

Confidential B2B SaaS client

Result

Onboarding: manual → self-serve

SaaS PlatformBackendCloud

Problem

The existing codebase had tenant logic scattered across the application layer, making it fragile to extend and risky to scale past a handful of customers.

Challenge

Rebuilding data isolation, permissions, and billing without a full rewrite — while the existing customers kept using the product.

Solution

Introduced a tenant-aware data layer, role-based access control, and a subscription/billing module, migrated incrementally behind feature flags with zero customer-facing downtime.

Result

The platform now onboards new organizations in minutes instead of requiring manual setup, with clean tenant isolation enforced at the data layer.

Vertical SaaS

Modernizing a legacy application without stopping the roadmap

Confidential product company

Result

Monthly → weekly release cadence

Web AppCloudModernization

Problem

Years of incremental changes had left the codebase tightly coupled, with no clear ownership boundaries and a deploy process the team dreaded.

Challenge

Modernizing the architecture while the business kept shipping — a full rewrite wasn't an option the team could afford.

Solution

Extracted the highest-risk modules into well-bounded services, introduced automated deployment and testing around them, and migrated the frontend incrementally, route by route.

Result

Deploy frequency increased and rollback risk dropped sharply, giving the team confidence to ship weekly instead of once a month.

Internal tools / SaaS

Adding fast search and real-time updates to a data-heavy product

Confidential SaaS client

Result

Search: seconds → <100ms

BackendSearchPerformance

Problem

Search queries hit the primary database directly, competing with production traffic and returning results only after several seconds.

Challenge

Adding real-time search and updates without adding operational complexity the team couldn't support long-term.

Solution

Introduced a dedicated search index with a sync pipeline keeping it up to date, and added caching for the read-heavy endpoints.

Result

Search response times moved from seconds to sub-100ms, and primary database load dropped enough to defer a planned infrastructure upgrade.

Have a similar problem?

If any of this looks familiar, let's talk about what's actually going on with your product.

Book a 30-minute call See our work