Work

I like work that leaves a team with something sturdier than the pitch.

The projects here span AI products, modernization work, and cloud-backed systems, but the pattern is usually the same: make the product clearer, make the system sturdier, and reduce the drag around it.

AI products that stay tied to reality

I build retrieval, workflow, and evaluation layers that help a product stay useful without pretending it knows more than it does.

Representative proof

TwinMail - AI-native email workflows

Offline .mbox ingestion supports archives up to about 55 GB.

Read case study Featured repo

Modernization that lowers support load

I like cleaning up high-friction workflows, reducing manual work, and leaving teams with fewer brittle steps to maintain.

Representative proof

EPIC operations modernization

Paper-heavy workflows were shifted into self-service systems.

Read case study

Cloud and data foundations that age well

The stack usually comes down to clear boundaries, reliable services, and infrastructure choices a team can keep living with.

Representative proof

TwinVault - secure AI-enhanced asset platform

Role-aware controls and audit-friendly lineage were built into the platform from the start.

Read case study Featured repo

Trust layer

Proof behind the delivery themes

If you want a feel for how I work, start with the public links, the repos, and the case studies. I would rather show the work than dress it up.

Verified links

LinkedInlinkedin.diplov.ai GitHub Orggithub.twindevs.ai/org GitHub Enterprisegithub.twindevs.ai/enterprise Xx.com/twindevsX YouTubewww.youtube.com/@twindevs

Featured repos

prod-DiplovThe portfolio, assistant, and intake surface behind this site.prod-TwinMailEmail workflow product work with retrieval, archive ingestion, and automation at the core.prod-TwinVaultSecure platform work around RBAC, lineage, and controlled AI-assisted access.prod-TwinSpacesWorkspace tooling that reflects the broader operating environment behind Twindevs.

Operating signals

Public links, case studies, and repos are here so you can verify the work yourself.

The assistant stays inside published material instead of improvising past the source.

Contact stays structured and protected so conversations do not disappear into a generic inbox.

The through-line

Start with the case studies if you want the story. Use the resume view if you want the experience framed for a role. Reach out if you already know what kind of help you need.

Next step

Want the short version?

Start a conversation Read case studies

Open assistant

The resume helps when you want the role-shaped version. The assistant helps when you want the quick version.