I’m Klim. I design, build, and scale AI systems from scratch — architecture, code, deployment, and everything in between. I like high-complexity, weird, or unsolved problems. If it’s AI and tough, I’ll figure it out, solo or with my own team of trusted devs. I don’t just design from a distance — I get hands-on, ship MVPs, and scale to production.
Key Achievements
- Developed NerdQA: Open-source research engine with focusing on statement transparency and novel reasoning backtracking via the proposed concept of citation trees. I published it 6 months before OpenAI’s Deep Research, which still doesn’t have this level of claim transparency
- Built an enterprise GenAI co-pilot: Natural language interface for 300,000+ APIs and 3,000+ database tables, all in Slack
- Pioneered early LLM adoption: Fine-tuned GPT-2 in 2019 for automated ideation, concept extraction, distillation and other tasks — way before it was cool
- Built autonomous on-chain AI systems: Agents that communicate with each other and make decisions directly on blockchain
- Architected web-scale data pipelines: Took a tiny database and scaled it to 250M+ docs with custom ML and NLP for trend analysis, concept extraction and advanced data enrichment
Core Capabilities
- AI System Architecture & Integration: Real-world AI designed to work with messy data, complex APIs, and legacy systems.
- Advanced RAG Systems: Knowledge graphs, hybrid search, and citation mechanisms that go way beyond the basics.
- LLM Application Development (Since GPT-2): Deep expertise from fine-tuning early models to engineering robust, optimized LLM-powered applications today.
- UX-Informed AI Design: Leveraging 8 years in UX to ensure AI systems solve real user problems effectively and are readily adopted.
- Multi-Agent Systems: I build agent frameworks that talk, plan, and get things done.
- Autonomous & Decentralized AI: Exploring and building self-governing, verifiable systems — on-chain and off.
- Practical AI Safety & Governance: Baking in technical safeguards, transparency, and control mechanisms from the start.
I like hard problems. The weirder, the better. I build, debug, and deliver — end-to-end. I can work solo (and love it), or bring in my own trusted team if you want to scale fast. No hand-holding required.
Core Principles
- Pragmatic Results: I ship real, working systems.
- Transparency & Trust: Everything I build is auditable and reliable.
- Lean Execution: Minimal overhead, max output.
- Research to Production: I turn new ideas into running code.
Upwork Profile
- 25+ completed projects with 5.0/5.0 average client satisfaction
- Top Rated Plus freelancer status with 100% Job Success Score
- Proven track record delivering complex projects remotely and asynchronously