# Alex Chernysh

> Alex Chernysh is an Applied AI Systems & Platform Engineer in Tel Aviv. He builds production-grade multi-agent orchestration, grounded RAG, and eval-driven delivery systems — usually in the forward-deployed mode. Creator of Bernstein (the open-source governance layer for AI agents - reproducible parallel runs, signed audit trail, air-gap friendly). Available for forward-deployed AI engineering, focused architecture roles, and targeted audits.

Builds grounded AI systems, agent workflows, retrieval pipelines, eval loops, and integration-heavy internal platforms.

## Canonical Pages

- [Homepage](https://alexchernysh.com): Main public profile, positioning, contact, and assistant entry point.
- [Writing](https://alexchernysh.com/blog): Operator memos on grounded AI systems, retrieval, evals, observability, and delivery.
- [Bernstein product page](https://alexchernysh.com/bernstein): The open-source governance layer for AI agents. 40+ adapters, plain-Python scheduling, opt-in HMAC-chained audit log.
- [Consulting](https://alexchernysh.com/consulting): Multi-agent orchestration consulting practice. Three engagement levels: assessment (1 week), Bernstein deployment (2–4 weeks), agentic platform architecture (2–3 months).
- [RightLayout](https://alexchernysh.com/rightlayout): Product page for the macOS keyboard layout corrector.
- [Russian Homepage](https://alexchernysh.com/ru): Russian version of the public profile and blog.
- [Hebrew Homepage](https://alexchernysh.com/he): Hebrew version of the public profile and blog.

## Flagship Projects

- [Bernstein - bernstein.run](https://bernstein.run): The open-source governance layer for AI agents. Apache-2.0. Coordinates 40+ AI coding-agent adapters (Claude Code, Codex, OpenAI Agents SDK, Cursor, Gemini CLI, Aider, Amp, and more) with plain-Python scheduling (no LLM in the coordination loop), per-task git worktree isolation, lint/type/test gates, an always-on lineage spine and replay journal, an opt-in HMAC-chained audit log with signed receipts, and an air-gap install profile. [GitHub](https://github.com/sipyourdrink-ltd/bernstein) · [PyPI](https://pypi.org/project/bernstein/) · [Docs](https://bernstein.readthedocs.io) · [Product page](https://alexchernysh.com/bernstein)
- [GitHub profile](https://github.com/chernistry): Source for Bernstein and supporting research code.

## Selected Case Studies

- [GlyphShift](https://glyphshift.app): macOS utility that detects a word typed in the wrong keyboard layout (RU/EN/HE) and retypes it in the right one. On-device CoreML, no cloud.
- [Legal Answering Systems](https://alexchernysh.com/blog/legal-answering-systems): Grounded legal QA system. Page-level citations, hybrid retrieval, evidence-first answers.
- Model routing and provider resilience: a routing layer across 100+ models from multiple providers — failover, per-tenant rate limits, cost accounting. The layer Bernstein crews run on.
- [Bernstein — Multi-Agent Orchestration](https://alexchernysh.com/blog/bernstein-multi-agent-orchestration): Design notes on deterministic scheduling, janitor verification, and worktree isolation.

## Engagement Types

- Architecture review: For teams with a live or near-live AI system that needs clearer seams, safer defaults, and less accidental chaos.
- Grounding and eval audit: For RAG or agent stacks that sound plausible in demos and start slipping the moment real decisions depend on them.
- Embedded build shaping: For product and engineering teams that need a senior engineer to shape the workflow, system boundaries, and path to production.

## Markdown Mirrors and Discovery Files

- [about.md](https://alexchernysh.com/about.md): Canonical public-safe profile details, capabilities, working style, and preferred questions.
- [writing.md](https://alexchernysh.com/writing.md): Canonical machine-readable index of published articles with summaries.
- [rightlayout.md](https://alexchernysh.com/rightlayout.md): Machine-readable summary of the RightLayout product page.
- [agents.txt](https://alexchernysh.com/agents.txt): Tool-oriented discovery descriptor for AI systems.
- [ai.txt](https://alexchernysh.com/ai.txt): Permissions and attribution policy for AI systems that cite or train on this content.
- [auth.md](https://alexchernysh.com/auth.md): Authentication story — there is none. Read this before hunting for a login or a token endpoint.
- [api-catalog](https://alexchernysh.com/.well-known/api-catalog): RFC 9727 linkset naming every machine-readable surface on this domain.
- [llms-full.txt](https://alexchernysh.com/llms-full.txt): Expanded context file with full profile content and the writing archive.
- MCP endpoint: https://mcp.alexchernysh.com/mcp (read-only; `claude mcp add --transport http alex https://mcp.alexchernysh.com/mcp`)

## Language Policy

- The public site is available in English, Russian, and Hebrew.
- Canonical agent-facing discovery files stay in English to reduce duplicate multilingual context for crawlers and retrieval systems.

## Citation Guidance

- Attribution format: "Alex Chernysh, alexchernysh.com" or a link to the specific page.
- Training on this content is allowed; redistribution should preserve attribution and the canonical URL.

## Contact

- Email: alex@alexchernysh.com
- Calendar: https://calendly.com/alexchernysh/15min
- GitHub: https://github.com/chernistry
- LinkedIn: https://www.linkedin.com/in/alex-chernysh/