My name is Stanislav Nabatnikov. I'm an AI Integration & Data Engineer.

I build code-first AI automation workflows, agentic pipelines, and resilient serverless ETL systems backed by a solid Data Engineering foundation on Google Cloud Platform.

These two featured projects are my personal sandboxes for semi-automated development. While I leverage automation to speed up routine setups, I personally design and integrate every new feature. Every decision—whether a success or a debugged mistake—serves as a practical learning ground, sharpening my skills in both development and Cloud Operations (Ops).

• Multifunctional cloud pipeline for Garmin FIT file processing
• Listener of telegram chats for certain keywords by personal account — serverless Telegram keyword monitor & alerting (GCF Gen 2): per-chat Firestore cursor state with at-least-once delivery, Secret Manager injection, heartbeat/health endpoints, 169-test offline pytest suite.

• AI & Agentic Workflows: Multi-agent architectures (CrewAI, OpenCode), LLM API integration (OpenAI, Claude, Mistral), RAG & Prompt Engineering.
• Data Engineering & Pipelines: Python (Asyncio/OOP), PostgreSQL/PostGIS, Firestore, ETL/ELT pipeline design, Idempotency, at-least-once delivery & cursor-based checkpointing, CDC patterns.
• Cloud & DevOps (GCP): Cloud Functions, Cloud Run, Pub/Sub, Secret Manager, Cloud Build, Cloud Scheduler, Cloud Monitoring alert policies, Docker/Podman, Linux (POSIX) system scripting, Bash automation.
• Technical Writing: Developer documentation, formal requirements & test-traceability documents (FR/NFR + RTM), architecture breakdowns, and knowledge base ownership.

• Documentation-driven development: Every solution I build is backed by structured, practical notes to streamline future maintenance and ensure seamless knowledge sharing within the team.
• Technical Writing & System Auditing: Passionate about analyzing software architecture and sharing findings with the global developer community. I write and publish analytical articles, breaking down architectural flaws and edge-case behaviors in popular platforms to advocate for better system design.
• Knowledge Base Ownership: Proactive in establishing and maintaining structured corporate wikis. I systematically analyze operational processes, implement improvements, and codify them into reusable documentation to ensure seamless team collaboration and onboarding.
• Custom Tooling & Scripting: Active builder of lightweight, specialized utilities to bridge gaps in existing software. I design custom micro-tools to automate manual tasks, control resource-heavy processes, and extend platforms with missing features.
  • 0-shredder with pauses for old HDD
  • Convert Gemini Web chats to Markdown for a knowledge base
• Modular Pipeline Design: Experienced in architecting decoupled, highly reusable codebases. I package core cloud integration components and boilerplate logic into standalone repositories to accelerate multi-project deployments and maintain a single source of truth.
  • Custom package for use with my GCP projects
• Production-Aligned Development: Operating in a native Linux (POSIX) environment, which guarantees 100% compatibility for containerized workflows (Docker/Podman) and automation scripts without the translation layers or hypervisors found on macOS/Windows.
  • Autonomous isolated code factory (5 roles) based on OpenCode
• System Automation: Skilled in leveraging native Linux tooling (systemd units, journalctl, bash scripting, and process signals) to build robust, predictable automation workflows
  • Install all necessary apps for clean Kubuntu
• Environment Replication: Experienced in turning step-by-step setup guides into fully automated, single-run Bash scripts parameterized with system environment variables for zero-friction environment provisioning.
  • Bash script for deploying the parent project at a clean CGP project from a personal Google account
• Pragmatic AI Engineering: Experienced in direct API integration with leading LLM providers. I prioritize pay-as-you-go API architectures to balance cost-efficiency with model performance, gaining a deep understanding of model limits, context-window constraints, and prompt formatting.
  • Local medical-consultative multi-agent pipeline on two independent models
  • Processing hybrid PDF files consisting of both digital PDFs and scans
• Operational Telemetry & Analytics: Designing systems around the philosophy that code is cheap, but execution data is priceless. I focus on building robust logging, tracing, and metric collection pipelines to turn raw application runtimes into valuable analytical assets.
  • Telegram keyword monitor: Firestore heartbeat, ?check/?health endpoints and Cloud Monitoring alert policy