Sergey K.

⚡ 7 years in software, 4 years building LLM products, and clients who stay 12+ months: from an ERP/CRM running ~120 businesses and 1,700+ work orders a day to multi-stage AI pipelines on AWS. 🗣 What clients say "He's easy to work with and genuinely cares about delivering results." — Charles Brun, Head of Sales, Entrepreneur (12+ months) "Very talented developer with a very positive attitude." — Rennen Chacham, CTO of Callers ($100k/month revenue) (12+ months) "Sergey's dedication, politeness, and exceptional expertise really stood out to me." — Rostislav Dugin, Founder & CTO of TgGroups (12+ months) 🛠 What I've built • Woodle Auto — cloud ERP/CRM for auto service businesses: bookings, work orders, inventory, sales, finance. ~120 businesses, 1,700+ work orders/day. Stack: JavaScript/Koa.js, React, PostgreSQL • Vyrex — AI creator discovery platform: scraping and data pipelines over thousands of YouTube/TikTok channels, with AI-generated metadata for semantic search. Maintaining in production for 8+ months. Stack: Claude Agents SDK, FastAPI, Celery, PostgreSQL, Next.js, AWS • Wizardly — real-time AI translation and voice narration: low-latency audio/video processing through multi-stage LLM workflows. Stack: OpenAI, FastAPI, Celery, AWS SQS, PostgreSQL, Next.js, Chrome extension • RangeTrue — healthcare AI lab-report analysis: PDF biomarker extraction, range scoring, and root-cause guidance. Stack: Anthropic Claude, LiteLLM, FastAPI, AWS Lambda, SQS, Aurora PostgreSQL, Terraform, Next.js • Askable — AI visibility platform: multi-LLM brand probing (ChatGPT, Claude, Gemini, DeepSeek), citation tracking, AEO scoring, weekly monitoring. Stack: AWS serverless, FastAPI, Next.js, Supabase 💡 How I can help • AI agents and multi-agent systems that automate real workflows: support, research, data entry, reporting, lead qualification • Full AI-powered products from architecture to production • RAG pipelines and LLM integrations (Claude, OpenAI, Gemini, LiteLLM, MCP) • FastAPI backends with async pipelines (Celery, SQS), plus React/Next.js frontends • Technical leadership for startups without a CTO: architecture, code review, team setup 🧠 My approach Most "AI automation" fails because it's built like a script, not a system. I treat every project like a product that has to survive in production: → Understand the workflow and business goal before writing code → Infrastructure that matches your stage: no AWS overkill for a 10-user MVP, no fragile VPS when you need to scale → Honest scoping and estimates, so you know what ships, when, and what it costs The goal isn't a clever demo. It's a system that runs unattended, handles edge cases, and actually saves you hours. AI Automation, AI Agents, Agentic AI, LLM Integration, OpenAI API, Claude API, Gemini API, RAG, Vector Database, AI Chatbot Development, AI Assistants, Workflow Automation, Business Process Automation, MCP (Model Context Protocol), Multi-Agent Systems, Prompt Engineering, Full Stack Development, Full Stack Developer, Web Application, SaaS Platform, MVP Development, React, Next.js, Node.js, TypeScript, REST API, API Integration, PostgreSQL