Multi-agent architectures, workflow automation, production-grade pipelines. No toy projects — real systems processing thousands of tasks daily, running 24/7, solving real problems.
|
Fully autonomous job search system. Aggregates 300+ jobs/day from 4 platforms, scores with rule-based + LLM two-stage pipeline, pushes qualified matches via Telegram. 8,700+ jobs processed. Runs 24/7 on a 15/month VPS.
|
|
Autonomous multi-agent content engine for Twitter/X. 4-layer voice validation, pattern learning with memory decay, slop detection, cost-optimized LLM routing with circuit breaker. 206 files, 39 API endpoints, 8 subsystems.
|
|
Intelligent model routing engine. Analyzes incoming requests, scores complexity, selects the cheapest capable LLM tier. Real-time token cost dashboard. 97% cost reduction vs. always routing to the largest model.
|
|
Upload PDFs, extract structured data, search a product catalog semantically via RAG, generate proposals. Works with any product database. Includes full demo mode without API keys.
|
|
Conversion-optimized quiz funnel engine. Conditional branching, multi-axis scoring, personalized PDF reports, email capture. Configurable via JSON. 40%+ quiz-to-lead conversion.
|
|
AI chatbot that qualifies leads through conversation, scores them, and routes qualified leads to CRM via webhooks. Integrates with Make.com, n8n, HubSpot, Slack.
|
|
Multi-agent autonomous sales pipeline. 4 specialized agents (Qualification, Outreach, Onboarding, Support) orchestrated via n8n, powered by Claude + Supabase pgvector RAG.
|
Languages
Frameworks & Libraries
AI / ML
Infrastructure
Automation & Tools
Dominik Tsatskin — Computer Science — Germany