Principal backend engineer · AI infrastructure architect
Principal backend engineer & AI infrastructure architect. 15+ years shipping production systems. Now building the infrastructure that lets AI agents actually do the work.
Vitalii Cherepanov
Principal Backend Engineer · AI Infrastructure Architect · Open Source Builder
The story
Started in 1998 writing Visual Basic for forestry enterprises. Spent the 2000s and 2010s shipping production PHP, Symfony, and Bitrix at every scale — fiscal payments, identity, e-commerce, ERP integrations. 100+ commercial Bitrix modules in production across the CIS.
Picked up Go in 2019, never put it down. Built OAuth2 / OIDC platforms, gRPC microservices, message-driven cores. Got tired of rebuilding the same observability primitives, so I codified them as ClearVibeArchitecture (CVA).
In late 2024, AI coding agents went from toy to tool. The bottleneck moved from the model to the surrounding infrastructure: memory, observability, multi-agent communication, statusbars that don't lie. So I started building it. total-agent-memory, claude-statusbar, braincore (getbraincore.com), a2abridge, phpfilebroker — open source, MCP-native, self-hosted.
Currently living and working in Serbia (Novi Sad), running independent contracts for clients in the US, Australia and Europe. Open to relocation for senior backend / AI-tooling roles, and to async remote work in the same direction.
Featured work
Production-grade infrastructure for the AI coding agent era. Self-hosted, MCP-native, MIT-licensed.
Distributed cognitive workforce runtime — the AISWARM + a2abridge + BrainCore triad.
Pure-Go orchestrator that runs Claude Code, Codex, DeepSeek and local Llama (LM Studio / llama.cpp) side-by-side on one codebase. Phase 6 closed: 33 internal packages, 8 binaries, distributed cognitive workforce runtime. KindPlanner expands tasks into 12 typed agent roles; auto_choose routes 30/30/20/20 across providers; a 3-tier verifier (command → regression-on-integration → LLM judge) gates merges; peer-helper consult (ASK_PEER) spawns cross-model helpers on demand; soft pause/resume protects dependency cascades; subtasks decomposed over MCP; tokens + cost tracked per provider. Workers run inside sandbox-exec / bwrap, each in its own tmux + git-worktree, with SQLite (modernc.org, no CGO) in ~/.aiswarm. Bubble Tea TUI and a Web UI with REST API show A2A and memory tabs live. Memory is a side-channel through MCP (total-agent-memory or BrainCore) — the runtime does not embed its own. Two open peers stand under it: a2abridge (A2A 1.0 mesh) and BrainCore (cognitive memory plane).
Persistent memory for Claude Code & Codex CLI.
Auto-extracted knowledge graph, multi-representation embeddings, and a 3D WebGL visualisation of every decision the agent has ever made. Self-hosted, Ollama-optional.
Local-first cognitive memory for AI coding agents.
Attention, decision graph, "the right to say I don't know" — MCP-native, plugs into Claude Code, Codex, Cursor, Cline. Pure-Go binary, zero CGO. getbraincore.com
Rich two-line status bar for Claude Code CLI.
Real-time model, context usage, tokens, cost, duration, git branch, cache stats — without leaving the terminal.
Open A2A 1.0 mesh for Claude Code, Codex, Cursor, Cline, Continue, Gemini.
Local directory daemon on 127.0.0.1:7777 speaking JSON-RPC + SSE + Push Notifications. Lets multiple coding agents broadcast contract changes and FYI without spam.
Filesystem as a message broker — exchanges, streams, dead letters.
Persistent streams, consumer groups, priority queues, DLQs, retries and metrics. Everything you expect from RabbitMQ, Kafka and NATS — but it's a single PHP package and a directory tree.
Bitrix Docker tooling that finally feels modern.
Repeatable, multi-stage Bitrix dev environments — PHP 8.x, MariaDB, push, sphinx, scheduled agents — all behind a single Makefile.
Six llama.cpp-inspired optimisation patterns, measured in PHP 8.4.
A reproducible benchmark suite that translates llama.cpp ideas — mmap'd weights, flat dense buffers, value pools, table dispatch, generator streaming, columnar layout — into idiomatic PHP 8.4 with JIT, plus a naive-vs-optimised case study importing 100K CSV rows into Postgres. Docker + Make, PHPStan level 8, no cooking the numbers.
Writing
Long-form pieces on memory, AI tooling, distributed systems and the boring backend work behind them.
Six llama.cpp optimisations translated to PHP 8.4 with JIT, benchmarked from 1M to 1B records. Half the hypotheses were wrong: SplFixedArray loses on speed, mmap is 7× slower per call, match equals switch. The other half become survival tools at scale.
Analysis of Anthropic's experiment running 16 parallel Claude instances to build a C compiler in Rust — what infrastructure had to be invented around the agents because the primitives weren't there yet.
Production AI should optimise for zero confidently-wrong actions, not accuracy percentage. Abstain as a first-class outcome, self-tasking, and a cognitive runtime are the path to trustworthy autonomy.
Atomic units with lifecycle, strict mode with abstention, causal decision chains, AST-based code identity, internal git, scoring mechanics, and negative memory with rule engines — the recipe behind real cross-session memory.
Vector search is search, not memory. Three failure modes — chunks lose meaning, no structural distinction between facts, and no temporal validity — make RAG hallucinate confidently without knowing what it doesn't know.
Patterns that quietly accumulate debt: phantom abstractions, copy-paste-by-prompt, silent error degradation, context amnesia between requests, and decorative tests. Human oversight of architecture and immediate refactoring are the antidote.
Total Agent Memory — an open-source MCP server providing persistent knowledge across sessions. 32 tools spanning storage, self-improvement via error tracking, knowledge graphs, episodic recall, and skill assessment.
Seven architectural principles distilled from etcd's Go codebase, mapped onto everyday PHP/Symfony work: typed contracts, single-responsibility services, composable middleware, observability as architecture, facade APIs, traceable request paths, and explicit dependencies.
An architectural style for backend systems that combines Hexagonal, Outbox/Inbox, Observability, mandatory Feature Toggles, and flow visualisation.
Immersive VR tool for debugging and analysing complex microservice flows with Go, gRPC, and Unity VR — bottlenecks in seconds, request flows in real time.
Reviews
“This is the first article on AI memory I've read that doesn't reduce it to vector search. Thank you.”
“Plugged total-agent-memory into our Codex setup. Recall went from 60 % to 94 % overnight. The MMR diversity flag is genius.”
“Vitalii ships open source the way most people ship internal tickets — with documentation, benchmarks, and zero excuses.”
“I've forwarded "RAG isn't memory" to every PM at our company. Should be required reading.”
“After two days of dogfooding braincore I cannot go back to a stateless agent. The "right to say I don't know" idea alone justified the install.”
Toolkit
Grouped by job, not by skill bar.
Now
BrainCore (getbraincore.com) — local-first cognitive memory for AI coding agents. Pure-Go, MCP-native, daemon.
total-agent-memory, claude-statusbar, a2abridge, phpfilebroker — six pinned OSS projects, all in active dogfooding.
Writing on Medium about AI tooling, memory architecture, and the infrastructure agents need to do real work.
Independent contracts with US / Australian / European teams. Open to relocation and senior AI-tooling roles.
Connect
Looking for Principal-level backend & AI infrastructure work — async remote contracts or on-site staff/principal roles. Open to relocation. Based in Novi Sad, working with US / AU / EU teams.