You will get an AI agent workflow that verifies itself and reports failures
Rising Talent
Project details
Most AI automations fail the same way: they run, they look fine, and they quietly skip things. No error, no alert - just missing rows, half-finished runs, and numbers nobody can trace back to a source.
I build agent pipelines the opposite way. Every run has a verifiable stopping condition: a check decides when the work is done, not a feeling. Failures are loud - errors and timeouts are retried and logged, completeness is reconciled against the source (so "the source had nothing" is never confused with "we dropped it"), and an interrupted run resumes from its cursor instead of starting over or double-writing.
What you get is not a demo that worked once. It is a workflow with a test that proves it, documentation your team can run without me, and honest limits written down - including what the system cannot do.
Typical uses: scheduled data collection and enrichment, document extraction with numbers traced to their source page, multi-step reporting loops, agent-to-agent handoffs that survive restarts.
Everything runs inside your own accounts - in most cases on free tiers, so your ongoing cost is zero.
I build agent pipelines the opposite way. Every run has a verifiable stopping condition: a check decides when the work is done, not a feeling. Failures are loud - errors and timeouts are retried and logged, completeness is reconciled against the source (so "the source had nothing" is never confused with "we dropped it"), and an interrupted run resumes from its cursor instead of starting over or double-writing.
What you get is not a demo that worked once. It is a workflow with a test that proves it, documentation your team can run without me, and honest limits written down - including what the system cannot do.
Typical uses: scheduled data collection and enrichment, document extraction with numbers traced to their source page, multi-step reporting loops, agent-to-agent handoffs that survive restarts.
Everything runs inside your own accounts - in most cases on free tiers, so your ongoing cost is zero.
AI Algorithms
Large Language Model, Transformer ModelAI Applications
AI-Generated CodeAI Development Language
PythonAI Models
GPT-4What's included
| Service Tiers |
Starter
$95
|
Standard
$220
|
Advanced
$420
|
|---|---|---|---|
| Delivery Time | 4 days | 7 days | 12 days |
AI Model Integration | - | - | - |
Batch Normalization | - | - | - |
Database Integration | - | - | - |
Detailed Code Comments | - | - | - |
Image Upscaling | - | - | - |
MLOps | - | - | - |
Model Deployment | - | - | - |
Model Documentation | - | - | - |
Model Monitoring | - | - | - |
Model Testing & Optimization | - | - | - |
Model Tuning | - | - | - |
Natural Language Processing | - | - | - |
NLP Tokenization | - | - | - |
Pre-Training | - | - | - |
Prompt Engineering | - | - | - |
Setup File | - | - | - |
Source Code | - | - | - |
1 review
(1)
(0)
(0)
(0)
(0)
This project doesn't have any reviews.
RT
Robin Paolo T.
Aug 19, 2026
Hikvision CCTV Plate Recognition Integration
Vasyl was excellent to work with from start to finish.
What impressed me most was that he did not simply try to sell us a solution. Before we even started the contract, he tested our actual CCTV footage, explained the limitations clearly, and proved that our existing cameras could work without us having to purchase new ANPR equipment.
Throughout the project, he consistently went beyond what was expected. He was patient, highly responsive, transparent whenever there was an issue, and persistent in troubleshooting until the system was working properly on our actual setup.
He successfully delivered a working ANPR system using our existing Hikvision CCTV cameras, including automatic plate recognition, duplicate/visit filtering, consolidated Excel reporting, repeat-customer tracking, and automatic operation on our office computer.
What I appreciated most was his attitude toward problems. Whenever we found an issue during testing, he did not make excuses. He investigated it, explained the cause clearly, made the necessary changes, and continued improving the system.
Very knowledgeable, dependable, and professional. He genuinely cared about whether the final system would be useful for our business, not just whether he could technically complete the job.
Highly recommended. We would gladly work with Vasyl again.
What impressed me most was that he did not simply try to sell us a solution. Before we even started the contract, he tested our actual CCTV footage, explained the limitations clearly, and proved that our existing cameras could work without us having to purchase new ANPR equipment.
Throughout the project, he consistently went beyond what was expected. He was patient, highly responsive, transparent whenever there was an issue, and persistent in troubleshooting until the system was working properly on our actual setup.
He successfully delivered a working ANPR system using our existing Hikvision CCTV cameras, including automatic plate recognition, duplicate/visit filtering, consolidated Excel reporting, repeat-customer tracking, and automatic operation on our office computer.
What I appreciated most was his attitude toward problems. Whenever we found an issue during testing, he did not make excuses. He investigated it, explained the cause clearly, made the necessary changes, and continued improving the system.
Very knowledgeable, dependable, and professional. He genuinely cared about whether the final system would be useful for our business, not just whether he could technically complete the job.
Highly recommended. We would gladly work with Vasyl again.
About Vasyl
AI Automation Engineer | Claude Code, n8n, AI Agents & Integrations
Lviv, Ukraine - 10:28 pm local time
4+ years in web development. The last 2 years I worked as a technical specialist inside high-volume performance-marketing teams, where speed and reliability directly equal money:
• 1,000+ landing pages built, adapted and deployed
• Full tracking infrastructure: Keitaro, AIO Tracker, S2S postback integration
• Servers, domains, DNS, SSL, CRM integrations — managed end to end
For the past year I've been building production software with Claude Code and multi-agent workflows at the core of my daily process. Real projects I've shipped:
✔ An AI-powered landing page generator for performance marketers — a live product that already generates revenue
✔ A production license-plate recognition system for a retail fuel business — 7 CCTV cameras (Hikvision), offline Windows deployment, installed remotely and running on-site today
✔ Several HTML5 browser games
✔ Telegram bots (automation, content delivery, user flows)
✔ A website builder / constructor
✔ A copy-trading & analytics tool for Polymarket
✔ Right now: deploying self-hosted Keitaro tracking infrastructure for a performance-marketing team, and building installable PWA apps
What this means for you: I deliver in days what usually takes weeks — and every project is tested and verified before handoff.
WHAT I CAN DO FOR YOU:
✔ Landing pages — pixel-perfect, responsive, fast-loading (HTML/CSS/JS, WordPress)
✔ Web apps & MVPs — from idea to working product
✔ Telegram bots & automation
✔ HTML5 browser games
✔ Tracking & affiliate infrastructure: Keitaro, AIO Tracker, postback (S2S) setup, servers, domains, CRM
✔ AI integration into your product or workflow (Claude API, chatbots, content pipelines)
AGENT-DRIVEN DELIVERY (how I actually build):
I design and run agent pipelines with verifiable stopping conditions — a check decides when work is done, not a feeling. The discipline is the point: every claim I make about a system is backed by a measurement I ran myself.
✔ Published a TypeScript MCP server for multi-agent shared memory and handoffs — 29 tests, green CI (repo linked in my portfolio below)
✔ I read every diff my tooling produces and catch the shortcuts it takes
✔ Systems I build report their own failures instead of failing silently — errors retried and logged, completeness reconciled against the source, resumable after interruption
HOW I WORK:
— Clear written communication, fast responses
— Small milestones, working demo early
— Everything is tested before delivery
Message me with what you need — you'll get a concrete plan and an honest estimate the same day.
Steps for completing your project
After purchasing the project, send requirements so Vasyl can start the project.
Delivery time starts when Vasyl receives requirements from you.
Vasyl works on your project following the steps below.
Revisions may occur after the delivery date.
Scope confirmed in writing
Scope in writing: I turn your description into a short spec - inputs, outputs, stopping condition - and you confirm it before I build.
Build with retries and logging
Data pulled with retries and full logging; results written idempotently, so a rerun cannot double-count anything.