Personal information hidden

Kostiantyn Pykhterov
AI/ML Engineer | AI Agents Developer

Summary
AI/ML Engineer specializing in intelligent agents for business platforms.
I build multi-agent systems using LLMs (like OpenAI, Claude, Gemini, etc.), combining
them with RAG pipelines that include vector databases (FAISS, Qdrant), knowledge
graphs (Neo4j), REST API calls, and high-quality embeddings to automate search,
reasoning, and decision-making.
My experience spans agentic frameworks (LangChain, LangGraph, LangSmith,
AutoGen, CrewAI, etc.), prompt engineering (System, User, Con g, Function calling,
Tools), and scalable backend development (FastAPI, SQLAlchemy).
I bring deep domain knowledge in ERP, logistics, insurance, retail, and
manufacturing — with a focus on solving real business problems through intelligent
automation.

Skills
AI & LLMs: LangChain, LangGraph, LangSmith, AutoGen, CrewAI, RAG, Prompt
Engineering
Embeddings & Vector DBs: Transformers, sentence-transformers, FAISS, Qdrant
Knowledge Representation: Neo4j
Backend: Python, FastAPI, SQLAlchemy, MCSR.
DevOps & Infra: AWS, Docker
Frontend, Mobile: Swift, SwiftUI. Familiar with Kotlin, Flutter, React, JavaScript.

Work History
01.05.2023 - now
AI/ML Engineer | AI Agents Developer
Learn and Solve, Learn & Solve
I have been actively designing and building AI agents to extend the capabilities of
business systems. These agents enhance traditional analytics with semantic search,
intelligent recommendations, and natural language interfaces.

The architecture leverages Retrieval-Augmented Generation (RAG) patterns,
combining vector databases (FAISS, Qdrant) with high-quality embeddings
generated using models from the transformers library. This allows LLMs to access
contextual enterprise data for more relevant and accurate responses. LLMs (OpenAI,
fi
Mistral) are orchestrated via LangChain and LangSmith to handle tasks like semantic
search, prompt generation, and decision-making.. LangGraph is used to structure
multi-step agent work ows with condition-based branching and memory, enabling
more controlled and deterministic behavior in complex decision ows.
The backends is built with Python (FastAPI, SQLAlchemy), and includes components
for working with relational data (PostgreSQL), knowledge graphs (Neo4j), and ML-
based scoring logic.
Together, these tools form a robust and scalable platform that supports smarter
operations for drivers, managers, and insurance analysts.

20.05.2021 - now
Senior Mobile Engineer | Team Lead
Learn and Solve, Learn & Solve
Mobile apps for ERP-driven transport and insurance platforms:
- Developed a mobile dashboard for eet managers to monitor vehicles and make
operational decisions.
- Built a driver app that tracks driving behavior and provides feedback to improve
safety and productivity.
- Created a mobile “virtual insurance boutique” that sells usage-based auto insurance
based on driving patterns.
- These apps were tightly integrated with ERP systems, laying the groundwork for cur-
rent AI agent solutions.

2015 – 2023
iOS Engineer — multiple companies
Worked on a variety of commercial applications, including:
– mobile ERP systems for service centers and transport eets
– taxi apps (for drivers and passengers) used by major providers in Eastern Europe
and NYC
– a usage-based insurance app that tracks driver behavior and adjusts policies
– an e-commerce platform and an educational app for children
– contributed to mobile apps used by truck drivers operating across the US and
Canada, focusing on safety, routing, and compliance integration.
fl
fl
fl
fl
 This product experience gave me deep insight into the logic and work ows of ERP
systems, mobility platforms, and digital insurance. Today, I use this domain knowl-
edge to build AI agents that optimize and automate these same operations. During
this period, I also gained solid experience as a frontend engineer — designing client-
facing logic, user ows, and ef cient data interactions in mobile environments.

Before moving into mobile development, I spent several years building ERP systems
for both retail and manufacturing companies. In retail, this meant working with com-
plex ows of goods: warehousing, procurement, inventory movement, sales, and ana-
lytics. In manufacturing, I worked on systems covering everything from raw material
planning and imports to production and distribution. Some of these systems also
supported online commerce — with advanced logistics, delivery optimization, and
supply chain analytics.
This practical experience gave me a strong grasp of enterprise data: both the kind
that can be queried with SQL, and the kind that requires deeper semantic modeling
or AI-driven analysis. It also helps me understand how real businesses think and op-
erate — a perspective I now bring to designing intelligent AI agents for ERP work ows.

Education
East Ukrainian State University. Physical and Mathematical faculty.
Teacher of physics, mathematics, computer science
Languages
English: Intermediate (B1- B2). Con dent in technical writing and professional com-
munication.
fl
fl
fi
fi
fl
fl