Oleksii Makeiev

Python AI Engineer

Geniusee • Full time

Apr 2026 - Present • 6 mos

Owned end-to-end delivery of 9+ production AI and automation systems across HR, Recruiting, Finance, and Sales, working with a high degree of autonomy - from requirements discovery and technical design through implementation, testing, and production support. These systems reduced manual processing time by up to 87.5% across measured processes, with several achieving payback in under 2 months. Key cases: ◉ Engineered a cascading invoice-matching system across 5 currencies with vendor-alias resolution - automated ~80% of invoice/receipt processing in a domain where matching errors carry direct financial risk. ◉ Designed and built a RAG-based AI customer support agent (LangChain, Qdrant, CRM) with a custom prompt-injection filter layer and intent-based escalation logic, automating order status, stock checks, and FAQ handling end-to-end. ◉ Redesigned an AI decision pipeline for employee reserve requests (n8n, LangChain, OpenAI), introducing strict, auditable decision logic based on Resource Manager criteria and separating deterministic business rules from LLM-generated analysis. Resolved historical-context contamination caused by a race condition and cut processing time by 50%, with a 1.2-month payback. Technologies: n8n, LangChain, OpenAI API, Anthropic Claude, JavaScript, Qdrant, Google Workspace API, Jira API, Make.com, KeyCRM, ManyChat.

Python Developer

N-iX • Full time

Nov 2023 - Apr 2026 • 2 yrs 5 mos

Worked on two projects: 1.Development of an AI assistant for a large international technology company. 2.Development of a turnkey e-commerce website for a foreign client. Project 1: Enterprise AI Agent for Global Tech Company ◉ Built a GPT-4o mini classification pipeline with 94.2% average model confidence, with 99.6% of tickets receiving confidence scores above 0.90, enabling fully automated routing and spam filtering. ◉ Designed an OpenAI → Anthropic failover mechanism to maintain service continuity during API outages and rate limits. ◉ Leveraged Polars for parallel preprocessing of large datasets within the classification pipeline. ◉ Implemented Hybrid Search (Qdrant + LlamaIndex) with RAGAS validation, achieving 81.6% context precision. Project 2: High-Performance E-commerce Ecosystem ◉ Reduced API response times to <90ms through Redis caching and PostgreSQL query optimization. ◉ Reduced checkout transaction errors by 22.4% by redesigning the cart module and adding Sentry monitoring. ◉ Optimized Django ORM, resulting in 40.1% faster catalog loading for high-traffic environments. ◉ Improved sprint velocity by 15% through more accurate task estimation and technical contribution during Scrum planning. Technologies: Python, FastAPI, Polars, OpenAI API, LlamaIndex, Qdrant, Structlog, Django REST Framework, PostgreSQL, Redis, Celery, Docker, Git, GitHub, GitLab CI/CD.