Understand
Physical workflow
Operators, instruments, constraints, and the real sequence of work.
Pavlo Barzdun Technical Operations & Automation
Practical systems for complex operations.
I build practical systems where real-world operations meet software, from embedded control and data workflows to reliable infrastructure. I use AI and ML when they solve a concrete operational problem.
Open to professional conversations & selected collaborationsWinnipeg, Manitoba, Canada
01 / About
The useful work happens between disciplines.
My path into technical systems started in operations, not in a software-only role. I learned to understand the real workflow first, then use the right mix of software, hardware, data, and process design to make it more dependable.
That lets me follow a problem beyond a single layer: from the operator and physical process, through instruments and control, into data, software, deployment, and recovery.
02 / Focus areas
Where I do my best work.
I work between the physical and digital sides of an operation: the people running the process, the instruments producing signals, the data explaining what happened, and the services that turn information into action.
Technical operations
Turn real-world technical processes into measurable, reliable, operator-friendly workflows.
Operational software & tooling
Build tools and services that reduce manual work, ambiguity, and operational drift.
Embedded systems & automation
Connect instruments, scripts, sensors, actuators, hardware, and people into usable systems.
Data, observability & applied AI
Use Python, SQL, telemetry, dashboards, practical ML methods, and stateful AI workflows to improve decisions and troubleshooting.
Infrastructure & reliable delivery
Run services, CI/CD, virtualization, backups, and deployment workflows with an emphasis on evidence and maintainability.
03 / Selected work
Selected systems and projects.
This selection follows the operating path from physical control and operational visibility to workflow platforms, applied AI tools, and focused public prototypes.
Primary operational systems
Systems that support real work.
Physical control, operational visibility, and software platforms shaped around the people and processes that use them.
01
Smart Pump Station
- Context
- A physical lab process with five independently controlled pump channels.
- Built
- Arduino control, PWM, Ethernet/HTTP communication, calibration data, remote actions, operating modes, and clear device status.
- Operating record
- The system remains in active production use as part of the operating workflow.
02
Sensor Station
- Context
- Environmental conditions need to be visible locally and useful as operational telemetry.
- Built
- An ESP32-S3 node combining sensing, a local status display, resilient Wi-Fi, InfluxDB telemetry, remote configuration, OTA update design, and wiring documentation.
- Ownership
- I own the specification, technical direction, review, and integration in an AI-assisted delivery workflow.
03
Operational Dashboards & Decision Support
- Context
- Operational data is useful only when metrics reflect the real workflow and help people decide what needs attention.
- Built
- Grafana dashboards, SQL-backed metrics, alerts, and reporting views for quality, throughput, and workflow visibility.
- Method
- I translate process rules into queries and visual signals, validate the interpretation, and maintain the surrounding data and delivery path.
04
Operational Workflow Platform
- Context
- Complex operational workflows need a consistent interface, explicit state, and service boundaries that can evolve safely.
- Built
- A modular web platform with authenticated workflows, queued background work, Microsoft Business Central API integration, controlled SMB file access, environment-specific delivery, and audit-friendly state.
- Ownership
- I shape the workflows, architecture, implementation review, testing, delivery, and recovery boundaries.
Applied AI & tooling
Tools that make agents and knowledge useful.
Working systems for controlled agent access, retrieval, local models, and evidence-backed delivery.
AI-01
MCP Workflow Control Server
AI agents need structured access to project work without receiving unrestricted write authority.
- Built
- A policy-governed Python Model Context Protocol server for structured project operations, with deterministic synthetic integration tests, diagnostics, and protected write paths.
- In practice
- I use it actively in day-to-day project delivery to inspect work, maintain task state, and preserve execution evidence.
AI-02
Local-first Knowledge Retrieval
Private conversation archives should be searchable without handing raw personal data to a third-party service.
- Built
- Hybrid search and citation-backed RAG using SQLite FTS5, embeddings, thread-aware context expansion, and local model backends.
- Privacy boundary
- Source archives stay read-only, while private content, databases, and logs remain outside version control.
Public prototypes
Focused builds, open for inspection.
Smaller public projects that show how I learn unfamiliar stacks through requirements, tests, review, and integration.
P-01
View projectOpen Lumina
A privacy-conscious native Apple-platform viewer concept for opening X-ray studies from ISO images and local folders, with a macOS-first architecture that keeps a later iOS version possible.
P-02
View projectCamera Capture
A focused public camera-capture project delivered in an unfamiliar Apple-platform stack through explicit requirements, tests, implementation review, and integration.
04 / Engineering approach
Measurable, maintainable, and boring in production.
Good technical software should reduce uncertainty, not create another layer of work. My bias is toward clear workflows, explicit ownership, useful logs, safe failure modes, and automation that removes repetitive work.
- 01
Understand the physical workflow before automating it.
- 02
Make state, ownership, logs, and failure modes visible.
- 03
Prefer maintainability and reversibility over cleverness.
- 04
Use testing and evidence to reduce operational risk.
- 05
Treat operators as part of the system design, not an afterthought.
- 06
Apply AI as an engineering tool, with human direction and review.
AI-assisted, human-directed
AI is part of the toolchain. Accountability stays with me.
AI is a working part of how I research, specify, prototype, test, review, and document systems, especially in unfamiliar stacks. I define the problem, constraints, and acceptance criteria; treat model output as a proposal until it passes source checks and executable tests; and retain control of integration, privacy, safety, and deployment. I also operate bounded agents, MCP tools, retrieval pipelines, and local-model workflows to understand their capabilities and failure modes in practice.
- Direction
- Problem framing, constraints, and acceptance criteria.
- Acceleration
- Research, prototyping, implementation, and documentation.
- Control
- Review, tests, privacy, integration, and deployment.
05 / Background
Operations discipline, built into the work.
My background spans technical operations, government administration, and tax operations. Across those settings, document control, reporting, audit support, and coordinated execution made accuracy and traceability non-negotiable.
Jul 2024 - present
Lab Technical Manager
Corvian
Technical management and practical systems work in a real lab operations environment.Dec 2022 - Jun 2024
Lab Operations Analyst
Corvian
Operations analysis, reporting, troubleshooting, process improvement, and workflow automation.2018 - 2022
Government operations, administration & tax roles
Document flow, audit support, controlled procedures, reporting, small-team coordination, and early workflow automation with Excel, VBA, Python, and API-derived data.06 / Contact
Working on a practical automation or real-world systems problem? Let's talk.
For collaborators and project contacts
I welcome thoughtful conversations about technical operations, automation, embedded systems, infrastructure, and applied AI/ML. I'm especially interested in exchanging ideas, contributing to open-source work, and discussing carefully scoped technical collaborations.
If you're reaching out about a project, tell me the technical or operational problem you're trying to solve and why you think my background may be relevant.