Built by one focused founder.
Deployed with you.
I'm Dennis - founder, builder, and systems architect behind Latence.
I work directly with teams to get retrieval infrastructure running in their environment.
Not for you. With you.
Small by design. Deep where it matters.
You work directly with the person designing the system, building the product, and making the technical trade-offs. That means faster decisions, less translation loss, and no consulting theater.
Direct accountability
The person you speak with is the one shaping the product.
Technical depth
Retrieval, model serving, and architecture stay in one set of hands.
Built together
The goal is a system your team can run, not a dependency to manage.
I help teams turn messy company knowledge into working enterprise AI systems.
The hard part is rarely the model. It is the layer beneath it: fragmented data, weak retrieval, governance, and deployment constraints.
Prove what works
Start on real company data, not demos. Identify what is technically and economically viable.
Design what scales
Shape retrieval, data, and deployment around real constraints.
Operationalize what matters
Turn technical requirements into something your team can actually run.
Built in production. Proven where failure is expensive.
From retrieval infrastructure to enterprise copilots, the work behind Latence has already been tested beyond demos.
Built the research, infrastructure, product, and commercial path from the ground up.
Delivered systems on real company data with traceable retrieval in regulated environments.
Built voyager-index and vllm-factory for modern multimodal, multi-vector retrieval.
Featured by HPE for work making enterprise AI more practical, efficient, and deployable.
Published ACM work on LLM training and system performance.
How I work with teams.
You bring the data, constraints, and environment. I bring the product, technical depth, and experience of getting retrieval systems running.
Start with the real problem
We begin on real data, not slides.
Work directly
Direct product and technical collaboration, without layers.
Build toward ownership
The handoff is a system your team can own and run.
If this is the kind of enterprise AI stack you want to build, let's talk.
Latence is for teams that care about retrieval quality, deployment control, and making AI useful on real company knowledge.
The waitlist is for the next enterprise release.