Skip to content
View Alexi5000's full-sized avatar
🎯
Focusing
🎯
Focusing
39 followers · 8 following

Achievements

Achievement: Pull Sharkx4Achievement: Pair Extraordinairex4Achievement: StarstruckAchievement: QuickdrawAchievement: YOLOAchievement: Public Sponsor

Achievements

Achievement: Pull Sharkx4Achievement: Pair Extraordinairex4Achievement: StarstruckAchievement: QuickdrawAchievement: YOLOAchievement: Public Sponsor

Highlights

  • Pro

Block or report Alexi5000

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Alexi5000/README.md

Alex Cinovoj

A terminal riding a tide: Alex. Build. Ship. TechTide AI.

AI & Infra Engineer. Building AI that survives outside the demo.

I build AI integrations, agent tooling, and workflow automation. My background is 13 years in US enterprise IT, including MSP work, systems engineering, and cloud infrastructure. That experience shapes how I approach automation: permissions, failure handling, and a clear path to recovery.

Website · LinkedIn · X · Substack

What I work on

  • AI agents and MCP integrations connected to real business systems.
  • Workflow automation with human approvals, logs, and recovery paths.
  • TypeScript, Python, Go, Rust, React, Tailwind, ShadCN, TanStack Start, Supabase, Azure, and cloud infrastructure.

Selected open source

  • Binnacle stars - Unified, zero-FFI, pure-Rust evaluator agent orchestrating the 15-tool TideOS evaluation stack across autonomous coding agents.
  • ClawKeeper stars - Agent control plane built around policy checks, spending controls, and audit logs.
  • CipherClaw stars - TypeScript toolkit for investigating multi-agent runs, tracing errors, and inspecting failure paths.
  • awesome-llm-apps stars - Audited agent templates for production readiness, dependency hygiene, and fail-closed error handling.
  • claude-code-templates stars - Hardened loopback listener authentication, credential redaction in diagnostics, and sanitization of pipe-to-shell installer flows.
  • 9router stars - Upstream routing patches and Cloudflare AI multi-modal reference image forwarding.
  • agent-lightning stars - Fixed memory leaks by pruning superseded routed experts across multi-turn trajectories.
  • vibe-kanban stars - Evaluated real-time multi-repo agent task board orchestration with Tauri and SPAKE2.
The TideOS 15-Tool Eval Stack (click to expand)
Step Engine Focus
01 reval stars Error analysis & causal Bayesian trace DAGs
02 gild stars Stratified golden dataset authoring & lockfiles
03 tessel stars Synthetic covering arrays ($P \times I \times C \times E$)
04 assay stars Deterministic assertions & pure-Rust DB/RLS
05 tide-rill stars Retrieval vs generation & faithfulness gating
06 keel stars Trajectory scoring & 14-step cost/latency bug detection
07 temper stars pass@k vs pass^k multi-trial reliability
08 parley stars Multi-turn simulated users & sycophancy tests
09 gavel stars LLM judge calibration & Rogan-Gladen debiasing
10 poise stars Double-blind order-swapped pairwise adjudication
11 plumb stars SplitMix64 bootstrap CIs & exact McNemar tests
12 weir stars Fail-closed per-category CI regression gating
13 eddy stars Online production sampling & CUSUM drift alerts
14 tare stars CPST (Cost Per Successful Task) & Pareto release gates
15 tether stars Failure harvesting, SimHash dedupe & release replay

How I work

I keep permissions narrow, make failures visible, and document how to operate what I build. I prefer a small system a team can maintain over a complicated one nobody owns.

Background & community

Founder & CTO, TechTide AI · CTO & Co-Founder, CPG Playbook

I also support developer communities as a Lovable Senior Champion and Ambassador, and a Discord moderator at TypeSafe AI.

Work with me · Email

Logs over vibes. Production over theater.

Contributing · Support · Security

Pinned Loading

  1. TechTideOhio/binnacle TechTideOhio/binnacle Public

    TideOS Eval Kit: The Evaluator Agent - Unified, zero-FFI, pure-Rust orchestrator for the 15-tool eval stack

    Rust 76 22

  2. ClawKeeper ClawKeeper Public

    Governed Claude agent control plane: deterministic policy engine runs before every LLM call, $0 default autonomous spend, database-enforced immutable audit log. 147 passing tests. MIT.

    TypeScript 29 7

  3. Shubhamsaboo/awesome-llm-apps Shubhamsaboo/awesome-llm-apps Public

    100+ AI Agents, Agent Skills and RAG Apps - Free and Open Source.

    Python 144k 21.2k

  4. CipherClaw CipherClaw Public

    Deterministic multi-agent debugger — zero LLM calls in the analysis path. Causal graphs, weighted root-cause scoring, failure prediction. 206 zero-mock tests, zero runtime deps.

    TypeScript 4 4

  5. TechTideOhio/techtide-harness-kit TechTideOhio/techtide-harness-kit Public

    Trust contracts for AI agent skills: JSON Schema risk tiers, tool permissions, deny-by-default egress, approval gates, prompt-injection fixtures, ~28 CI validation gates. Apache-2.0.

    Python 1

  6. TechTideOhio/swarm-357 TechTideOhio/swarm-357 Public

    357 Claude agent roles across 6 business layers. Portable Memvid memory, enforced cost caps, bash policy gate, durable checkpoints. Apache-2.0 Python runtime.

    Python 2 4