Independent AI engineering · Kent, UK
I'm Vadim, the engineer behind AI Church. I design and build AI agents, retrieval (RAG), and the cloud infrastructure to run them — taking products from prototype to production, with a bit of frontend to tie it together.
20+ years building large-scale, high-load systems · the last decade on ML products
Start here
Most teams lose hours every week to work that doesn't need a person — copying data between tools, chasing the same updates, answering near-identical emails, turning notes into reports. You already know which tasks they are.
You don't need to understand AI or write a brief. Just describe one routine the way you'd explain it to a new hire — in your own words, however messy.
One workflow, in plain text. Two lines or two paragraphs — whatever's easy.
I come back with what's automatable, how I'd build it, roughly how long, and what it'd cost.
You approve the plan, I ship it, and you get your time back. No lock-in.
Free, no obligation. Prefer email? hello@ai.church
End-to-end AI — from agents and retrieval to the infrastructure that runs them.
Autonomous and multi-agent systems that research, reason, run code, and act — with guardrails for cost and human oversight.
Retrieval and long-term memory that ground LLMs in your data — embeddings, entity graphs, and provenance you can trust.
Production cloud backends — serverless on GCP, async pipelines, multi-tenant APIs, auth, billing, and observability.
Prototype to shipped product, including the frontend that ties it together — design, build, deploy, and evaluate.
Based on the Kent coast and open to local work. Beyond the products below, I help local businesses automate the repetitive, time-eating parts of their day and adopt AI sensibly — an enquiry or booking assistant, document and data automation, a website, or just honest advice on what's actually worth doing. Plain English, no enterprise price tag.
Talk about your projectBehind AI Church is one senior engineer — not an agency. Two decades building large-scale systems (search ranking, recommendations, and trust-and-safety ML at global scale), the last ten years focused on machine-learning products. You work directly with the person writing the code.
Production AI products I've designed and built.
An AI relationship coach for couples, delivered through both a Telegram bot and an iOS app on one cloud backend. Structured long-term memory keeps context across conversations, and a privacy architecture is engineered so one partner's confidences never leak to the other.
A live gateway connecting AI assistants like Claude and ChatGPT to real accounts — Telegram, banking — over one OAuth-secured endpoint, with per-user encryption.
Visit projectA proactive AI project manager that lives in team chats, builds an auditable memory of people, projects and decisions, and nudges the right person before deadlines slip.
Residents report local issues to a conversational assistant that routes each case to the right councillor, who gets AI-drafted replies grounded in their own approved answers.
Visit projectVersion prompts, build datasets, and grade model outputs by exact-match and an AI judge across OpenAI, Anthropic and Google — with cost and latency tracking.
Visit projectShipped products I helped engineer.
An AI skincare assistant that scans your face and any product to score it on safety, cost and efficacy, then recommends personalized routines from a 500K+ product database.
Visit projectAn algorithmic news aggregator that clusters coverage of the same event from across the media spectrum, so readers can compare how outlets framed a story and judge for themselves.
Visit projectTell me about your project — local automation or a full AI build.