Artem S.

AI engineer who ships LLM apps, RAG, AI agents and voice AI that cut manual work and hold up with real users and real data, not just in a demo.

▪ Top Rated Plus | 100% Job Success | $100K+ earned on Upwork across 9 jobs | 3,800+ hours logged
▪ AI, Python and AWS

★★★★★ "Artem was a pleasure to work with. He approached the project with professionalism, delivered high-quality results, and maintained clear and prompt communication. His technical expertise and reliability exceeded expectations."
★★★★★ "They spun up a prototype for me in a super quick timeframe on very short notice." (Python + OpenAI API project)

◆ WHAT I BUILD
→ RAG pipelines and AI chatbots
An LLM chatbot is only as good as what it is allowed to read. I build RAG over your documents, tickets, product data and internal wikis: chunking, embeddings, retrieval and evaluation, so every answer comes back with its source instead of a confident guess. Send me the questions your own team actually asks and I will test retrieval against them before anything goes in front of a customer.

→ AI agents and automation
Automation that runs without anyone watching: intake, routing, data enrichment, document processing, reporting, and the handoffs between the tools you already pay for. When a step needs judgment, it becomes an AI agent: the model suggests the next action, and fixed rules decide whether it may run. Each agent has defined tools, defined limits and an action log you can audit. If a step touches money, customers or production data, a person approves it first.

→ Voice AI and call-center AI
For a retail call center I built a voice emotion analysis system: I benchmarked CRNN, WavLM and LCNN models on real call recordings, then shipped a CPU-optimized inference API in Python, FastAPI and PyTorch. Operators get live prompts and escalation alerts, and supervisors see each call's dynamics on a dashboard. I also built an AI chatbot with ElevenLabs API integration for a web platform.

→ Python development and API work
Plain, readable Python your own team can still read six months later. Django, FastAPI and Flask services, custom REST and GraphQL APIs, data scraping, ETL and process automation scripts, and integrations with the OpenAI API, Stripe, Twilio and whatever else your stack runs on. I am an AI engineer who ships LLM apps, not a prompt writer.

◆ FROM PROTOTYPE TO PRODUCTION
Many teams arrive with a demo that works on five examples and breaks on real users, real data and real load. I rebuild the weak parts: error handling, retries, authentication, logging, AWS deployment and tests against real cases. Send me the repository or a demo link and tell me what breaks first.