Browse categories
Explore
Fiverr Pro
English
$
USD
I will audit and harden your n8n ai agent or API workflow
Onur K
About this gig
AI automation can work in a demo and fail with real data, retries, rate limits, integrations, and human decisions.
I audit and harden n8n, Make, AI-agent, webhook, API, CRM, database, and voice-AI workflows. I find duplicates, silent failures, missing retries, weak state, race conditions, brittle integrations, unclear permissions, high API/LLM costs, and missing human escalation.
Evidence-led, I map data and state, trace failure paths, prioritize risk, and implement bounded package fixes. I validate idempotency, webhooks, logging, observability, error handling, and handoff documentation.
I build and operate customized agent infrastructure using the open-source Nous Research Hermes Agent runtime. Across its operating history: 1000+ scheduler entries, 5000+ scheduled LLM-backed jobs, 99+ configured MCP servers, and 50+ integrated capabilities, with persistent state, human checkpoints, observability, and fail-closed controls.
You receive a clear report, tested changes when included, and handoff notes. Full rebuilds, 24/7 monitoring, vendor fees, and guarantees are outside scope.
Send sanitized exports, screenshots, logs, and constraints. Never s
Get to know Onur K
Onur K
AI Automation and Agent Systems Engineer n8n APIs MCP Voice AI
- FromUnited States
- Member sinceMay 2026
- Avg. response time1 hour
Languages
English, Turkish
AI automation and agent-systems engineer focused on reliable workflows, APIs, persistent state, and human handoffs.
Builder of KAVI/Hermes, a multi-agent runtime with 152 scheduler entries, 25 scheduled LLM-backed jobs, 10 configured MCP servers, observability, human checkpoints, and canary/rollback-aware controls.
I work across n8n, Make, REST APIs, webhooks, Python, FastAPI/Flask, Supabase/PostgreSQL, RAG, MCP, Twilio, ElevenLabs, CRM automation, and workflow architecture. Hands-on with generative AI since 2022.
Also 10+ years in customer-facing sales, hospitality, and operations.
My Portfolio
Other AI Development Services I Offer
FAQ
What platforms do you work with?
I work with n8n, Make, custom REST APIs and webhooks, CRMs, Supabase/Postgres, Twilio, ElevenLabs, and OpenAI or other LLM APIs. I prefer sanitized exports or read-only access.
Can you fix an existing n8n workflow?
Yes. Standard and Premium include bounded repairs within the agreed scope. Please provide the workflow export, error examples, and expected behavior. Never send passwords or API keys.
Can you work with custom APIs and webhooks?
Yes. I can review authentication boundaries, payloads, rate limits, retries, idempotency, data mapping, and failure handling. You keep all credentials in your own environment.
Can you rebuild my entire automation?
Not within these packages. They cover audits and clearly bounded repairs or hardening. A full rebuild requires a separate scope, timeline, and quote.
Do you work with AI agents?
Yes. I can review tool permissions, prompts-to-actions, memory and state, model failures, retries, cost controls, and human escalation. I do not promise unrestricted autonomous decisions.
What access will you need?
Start with sanitized exports, screenshots, logs, architecture notes, and expected behavior. Read-only or temporary scoped access is preferred when necessary.
What if my system needs more work than the package includes?
I will document out-of-scope findings and recommend the next step. No additional implementation or production change is performed without a new agreed scope.
Will you make changes directly in production?
Not by default. I prefer staging, test copies, or controlled scoped access. Any production change must be explicitly agreed in scope, have an appropriate rollback path, and stay within the package boundaries.
