01
AI that stays tied to the source
I build retrieval and workflow systems that stay close to the material they came from, so the product is useful without pretending to know more than it does.
Dylan Dahal
I do my best work where product urgency and engineering discipline have to coexist.
I do my best work where product urgency and engineering discipline have to coexist.
Most of the work here sits somewhere between AI, platform architecture, and modernization. I like clear systems, calm rollouts, and tools people can trust after the launch buzz is gone.
How I work
Founder context, hands-on engineering, and a bias toward calm systems.
Twindevs is the studio behind the work, but this site is still personal. It is the clearest way to see how I think, what I build, and what actually shipped.
Entity
Twindevs, LLC
Primary host
dossier.diplov.ai
Delivery mode
Founder-led with hands-on build involvement
02
Modernization without the drama
I like untangling brittle processes, replacing manual handoffs, and leaving teams with something easier to run than what they started with.
03
Cloud systems people can live with
The architecture work tends to focus on reliability, access control, maintainability, and the boring decisions that matter once a system is real.
Trust layer
What you can verify here
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.
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.
Featured case studies
Published work with enough detail to feel the architecture underneath it.
AI-native workflows01
Key impact
Offline archive ingestion up to about 55 GB
Founder and product architect
TwinMail - AI-native email workflows
ML-native email productivity workflows with background categorization, summarization, and offline archive ingestion.
Proof signal
Offline .mbox ingestion supports archives up to about 55 GB.
Secure data platform02
Key impact
Role-aware asset and credential controls
Founder and platform architect
TwinVault - secure AI-enhanced asset platform
Secure storage and management for digital assets and credentials with privacy-focused architecture.
Proof signal
Role-aware controls and audit-friendly lineage were built into the platform from the start.
Operations modernization03
Key impact
Paper-heavy workflows shifted into self-service systems
Operations modernization lead
EPIC operations modernization
Modernized legacy, paper-heavy operations with practical automation and self-service systems.
Proof signal
Paper-heavy workflows were shifted into self-service systems.