Sheldon Glowatski | Vyreon Labs

Mechanical Engineer | APEGA Engineer-In-Training | Telemetry, Data, And Probabilistic Systems

Calgary, Alberta, Canada

LinkedIn · Research · Vyreon Labs · Contact

Sheldon Glowatski is a mechanical engineer and APEGA Engineer-in-Training who designs, deploys, and operates measurement, analytical, and software systems for noisy, uncertain, and partially observable environments.

His background spans real-time downhole telemetry, safety-critical field operations, production engineering, engineering automation, signal processing, time-series modeling, probabilistic state estimation, data validation, backend software, and cloud infrastructure.

He is the developer and operator of Vyreon Labs, an independent quantitative research and systems engineering project focused on financial telemetry. Vyreon converts options-market and price data into multi-horizon expectation states, uncertainty estimates, volatility diagnostics, structural measurements, and recurring public reports.

The common thread across this work is not drilling or financial markets by themselves. It is the engineering problem underneath both:

How do you extract reliable information from noisy measurements, preserve causality, quantify uncertainty, detect failure, and make the result useful to a person making decisions?

At A Glance

Professional registration: Engineer-in-Training with the Association of Professional Engineers and Geoscientists of Alberta.
Education: Bachelor of Science in Mechanical Engineering, University of Calgary, graduated with distinction, GPA 3.7.
Current work: Developer and operator of Vyreon Labs, a live financial telemetry, probabilistic state-estimation, validation, and reporting platform.
Industrial background: More than six years supporting real-time telemetry and field operations with Halliburton Sperry Drilling, plus a 16-month production and exploitation engineering placement with Canadian Natural Resources Limited.
Core technical areas: Python, SQL, PostgreSQL, FastAPI, Linux, Git, data pipelines, signal processing, Kalman filtering, model calibration, anomaly detection, and technical documentation.
Location: Calgary, Alberta, Canada.

Selected Engineering Outcomes

Designed, deployed, and operates an end-to-end quantitative telemetry platform spanning data ingestion, causal feature processing, probabilistic estimation, validation, storage, APIs, infrastructure, and public reporting.
Built and live-tested a real-time market bifurcation detector using continuous wavelet transforms, higher-order derivative signatures, template matching, and a lightweight convolutional neural network.
Identified anomalous high-temperature magnetic-ranging measurements that enabled correction of approximately 30 metres of wellbore placement error.
Reduced electromagnetic telemetry battery consumption from 54 amp-hours to 31 amp-hours per lateral, approximately 43 percent, while eliminating battery-related nonproductive time.
Isolated cathodic-protection interference during brownfield SAGD operations and restored magnetic-ranging and electromagnetic telemetry performance.
Automated recurring production forecasting and engineering-reporting workflows at CNRL, reducing manual processing effort by more than 90 percent.
Designed and programmed the C++ control architecture for an autonomous vertical-takeoff-and-landing aircraft that achieved stable flight and received the Hunter Hub Best Prototype award.

Current Work: Vyreon Labs

Sheldon is the developer and operator of Vyreon Labs.

Vyreon began as an independent market-systems research project and developed into a production financial telemetry platform. The public system currently focuses on SPY and operates at a daily cadence. It publishes recurring reports that describe options-market structure, forward expectation states, uncertainty, innovation-based volatility, and model calibration across several time horizons.

The current production architecture separates the options matrix into four maturity buckets:

Near-term, 8 to 30 calendar days.
Short-term, 31 to 60 calendar days.
Medium-term, 61 to 120 calendar days.
Long-term, 121 to 365 calendar days.

Each horizon maintains its own feature state, learned relationship, adaptive coefficient state, expected return, and uncertainty estimate. The system preserves disagreement among horizons rather than forcing them into one simplified directional label.

The production platform includes:

Daily options-chain and underlying-price ingestion.
Contract filtering and data-quality checks.
Options-structure, volatility-surface, sensitivity-field, and price-context features.
Causal endpoint Savitzky-Golay processing.
Exponential smoothing and causal exponential normalization.
Horizon-averaged forward-return targets and delayed target maturation.
A causal rolling return baseline.
Regularized linear modeling.
Recursive coefficient-state estimation through adaptive Kalman regression.
Predictive uncertainty and innovation-variance tracking.
Coverage, error, bias, drift, and volatility-signal validation.
PostgreSQL storage, read-only analytical services, and Linux infrastructure.
Automated weekly reporting and public web publishing.

The platform is operated as a real production system, not as a one-time notebook result. Routine work includes investigating data gaps, checking feature definitions, tracing model-state changes, testing releases, rebuilding causal warmup states, verifying report continuity, monitoring infrastructure, and correcting assumptions when an output looks implausibly strong or inconsistent.

The current public model should be understood as a measurement and state-estimation system. It does not claim to know one exact future price. It publishes a central estimate, an expected range, structural context, and evidence about whether the estimator remains aligned with realized behavior.

Read the current methodology
Explore the research program
View the SPY report archive
Learn how to read the charts

Engineering Approach

Measurement Before Interpretation

Sheldon separates numerical measurement from narrative interpretation.

The model produces the source data first. Written reports are generated afterward from structured measurements. A Mixed state is allowed to remain Mixed. A wide interval is reported as wide. A result is not translated into a stronger claim merely because stronger language would be more marketable.

Causality As A Hard Requirement

A historical estimate must use only information available at that point in time.

This requirement applies to signal processing, normalization, rolling baselines, target maturation, model prediction, state updates, uncertainty scoring, and validation joins. When causal ordering is wrong, the output is wrong even when the chart looks convincing.

Validation Separate From Marketing

A high calibration number is not automatically a good result.

Coverage must be interpreted alongside interval width, average error, visible bias, maturation delay, and the exact scoring convention. Adaptive filter diagnostics and strict issue-date forecast validation answer related but different questions. They should not be presented as though they are interchangeable.

Production Behavior Over Visual Appeal

Sheldon places more weight on systems that remain stable, diagnosable, and observable than on models that produce one impressive chart.

This means monitoring failure states, preserving logs, defining data-quality checks, testing recovery behavior, documenting assumptions, and designing outputs that can be audited later.

Failure As Engineering Evidence

Vyreon's research archive preserves rejected hypotheses, broken implementations, weak walk-forward results, and live deployment failures.

The purpose is not to celebrate failure. It is to retain the evidence required to understand why an approach changed and what knowledge survived.

Read the research history and selected case studies

Professional Experience

Halliburton, Sperry Drilling

MLWD Field Professional II, SAGD Operations
2012 To 2016 And 2022 To 2026
Western Canada, United States, and Oman

MLWD means measurement while drilling and logging while drilling.

Sheldon supported real-time downhole telemetry, survey integrity, well-placement certainty, magnetic ranging, anti-collision operations, formation-evaluation systems, and surface acquisition equipment in steam-assisted gravity drainage and other drilling environments.

The work required continuous interpretation of noisy and sometimes conflicting sensor data. Measurements could be delayed, distorted, degraded by interference, or affected by temperature and surrounding infrastructure. The operational decision was rarely based on one channel alone.

Responsibilities included:

Monitoring live sensor, survey, and telemetry data.
Configuring downhole tools and surface acquisition systems.
Verifying communication channels, sensors, and equipment readiness.
Diagnosing signal loss, telemetry degradation, tool faults, and survey anomalies.
Supporting electromagnetic and pulse telemetry systems.
Integrating electromagnetic repeaters with directional-drilling systems.
Leading magnetic-ranging and anti-collision operations.
Coordinating operators, directional drillers, geoscientists, remote technical personnel, and field teams.
Conducting structured failure investigations and documenting corrective actions.
Exercising stop-work authority when measurement or safety thresholds were not met.

Selected Field Investigations And Improvements

High-Temperature Magnetic-Ranging Anomaly

Sheldon identified anomalous ranging measurements under high-temperature conditions. The investigation enabled correction of approximately 30 metres of wellbore placement error.

The work required distinguishing a measurement problem from an actual positional deviation, comparing independent information, communicating uncertainty, and coordinating a correction without treating one instrument as unquestionable truth.

Electromagnetic Telemetry Optimization

Sheldon optimized QPSK-based electromagnetic telemetry through field testing, configuration changes, and performance analysis. EM 3DD battery consumption was reduced from 54 amp-hours to 31 amp-hours per lateral, approximately 43 percent, and battery-related nonproductive time was eliminated.

Cathodic-Protection Interference

During brownfield SAGD operations, Sheldon isolated cathodic-protection infrastructure as a source of interference affecting magnetic-ranging and electromagnetic telemetry. Systematic testing and configuration changes restored signal performance.

Telemetry Reliability In Weyburn

Through structured signal optimization and troubleshooting, Sheldon reduced MWD signal-related nonproductive time to zero during the referenced Weyburn operations.

These examples involve the same underlying disciplines that later shaped Vyreon: signal interpretation, sensor validation, anomaly detection, root-cause analysis, state estimation under uncertainty, operational risk control, cross-functional communication, and verification after corrective action.

Canadian Natural Resources Limited

Production Engineering And Automation Intern
2020 To 2021, 16-Month Placement
Calgary, Alberta

Sheldon supported the Jackfish SAGD production engineering team and the Deep Basin North production and exploitation engineering teams.

His work included production surveillance, decline-curve analysis, forecasting, investigation of unexpected well behavior, integration of operational datasets, engineering reporting, requirements gathering, and validation of analytical outputs against source data.

He developed Python, SQL, and Excel/VBA workflows that automated recurring production forecasting and engineering-reporting processes. The resulting tools reduced manual processing effort by more than 90 percent.

This work reinforced a principle that later became central to Vyreon:

A useful model is not complete until the surrounding data definitions, validation checks, interfaces, documentation, and operating workflow are reliable.

Selected Engineering Project

Autonomous VTOL Aircraft

Controls And Systems Integration, Mechanical Engineering Capstone
University of Calgary
Hunter Hub Best Prototype Award

Sheldon designed and programmed the C++ control architecture for an autonomous vertical-takeoff-and-landing aircraft using the open-source ArduPilot platform as a base.

The project integrated sensors, embedded hardware, actuators, control logic, and mechanical systems. The team developed the aircraft through repeated testing, troubleshooting, and multidisciplinary integration until it achieved stable autonomous flight.

The project established several engineering habits that continue to shape his work:

Timing and interfaces matter as much as individual components.
A controller must operate inside a complete physical system.
Failures must be reproduced and observed.
Stable operation is more important than a one-time demonstration.
Mechanical, electrical, and software definitions must remain consistent.

Technical Capabilities

Software And Data Systems

Python, Pandas, NumPy, SQL, PostgreSQL, FastAPI, REST APIs, Linux, Docker, Git, GitHub, Nginx, cloud deployment, ETL pipelines, data integration, automated reporting, data-quality validation, monitoring, observability, fault detection, backend troubleshooting, technical documentation, and release management.

Modeling, Signal Processing, And Validation

Time-series modeling, probabilistic state estimation, adaptive Kalman regression, ridge regression, predictive uncertainty, calibration, innovation and residual analysis, drift monitoring, causal Savitzky-Golay filtering, exponential smoothing, adaptive normalization, continuous and discrete wavelet transforms, multiscale feature extraction, higher-order derivative signatures, convolutional and recurrent neural networks, walk-forward validation, out-of-sample testing, anomaly detection, bifurcation detection, and model failure analysis.

Industrial And Engineering Systems

Real-time telemetry, MWD and LWD systems, downhole and surface instrumentation, electromagnetic and pulse communications, magnetic ranging, anti-collision, SCADA exposure, PLC programming, Structured Text, Ladder Logic, root-cause investigation, equipment diagnostics, reliability, operational risk, engineering calculations, technical drawing review, Autodesk Inventor, and SolidWorks.

Technical Communication And Evaluation

Requirements, procedures, investigation records, training material, model documentation, public technical reporting, causal and statistical review, and evaluation of AI-generated technical and software outputs for hallucinations, reasoning defects, and specification failures.

What Sheldon Brings To An Engineering Team

End-To-End Ownership

Sheldon is comfortable working from problem definition through deployment and ongoing operation. That includes gathering requirements, building data pipelines, constructing analytical logic, validating outputs, deploying services, diagnosing failures, documenting changes, and maintaining the resulting system.

Field Judgment And Software Depth

His background combines physical operations with software and data systems. He understands that a clean database value can still originate from a bad sensor, that a valid model can still be used incorrectly, and that production failures often occur at interfaces rather than inside the most sophisticated algorithm.

Work Under Uncertainty

Both field telemetry and market telemetry involve partial observation. Sheldon is accustomed to comparing imperfect measurements, checking independent evidence, preserving safety margins, communicating what is known and unknown, and changing course when the system does not support the original conclusion.

Failure Investigation

Unexpected behavior is treated as an investigation problem. The typical process is to reproduce the issue, verify definitions and timing, inspect data lineage, isolate the failing layer, correct the cause, rebuild affected state, and confirm recovery through downstream outputs.

Communication Across Disciplines

Sheldon has worked with operators, directional drillers, geoscientists, production engineers, exploitation engineers, remote technical specialists, software systems, and public readers. He can move between field language, engineering analysis, code, model behavior, and plain-language documentation without pretending those are the same discipline.

Education And Professional Registration

Bachelor Of Science In Mechanical Engineering
University of Calgary
Graduated with distinction, GPA 3.7

Engineer-In-Training
Association of Professional Engineers and Geoscientists of Alberta

Relevant academic and project work included control systems, system modeling, embedded programming, PLC programming, multidisciplinary design, mechanical systems, instrumentation, and autonomous controls.

Selected Public Work

Proprietary Vyreon source code, trained artifacts, coefficients, operational thresholds, and internal diagnostics are not public. Public repositories should be listed individually as they become available.

Editorial Responsibility And Use Of Automation

Sheldon is responsible for the public content published by Vyreon Labs.

This includes reviewing reports, maintaining methodology and educational pages, checking technical claims against current production code, correcting material errors, and distinguishing numerical output from written interpretation.

Automated tools, including language models, may assist with organization, grammar, translation-friendly wording, and conversion of structured model output into readable prose.

They do not generate the underlying market measurements.

The quantitative system performs data ingestion, feature processing, state estimation, uncertainty calculation, target maturation, and validation. Sheldon reviews the final public content and remains accountable for what is published.

Professional Scope

Sheldon's Engineer-in-Training registration is an engineering credential.

It is not a securities, investment-advisory, brokerage, accounting, or portfolio-management credential.

Vyreon's public reports are independent quantitative research. They do not consider an individual reader's financial circumstances, objectives, tax situation, liquidity needs, or risk tolerance. They do not provide personalized investment advice.

The relevance of Sheldon's engineering background is methodological. It explains experience with telemetry, uncertain measurement, validation, production systems, and failure investigation. It does not replace the need for the financial system's claims to stand on their own data, methodology, limitations, and observed calibration.

Read the full disclaimer

Independence, Conflicts, And Corrections

Vyreon Labs is independently developed and operated.

The public reports are not written on behalf of a broker, exchange, bank, fund, issuer, or financial news organization.

Sheldon may hold positions in securities or derivatives related to markets discussed by the site. A material position or commercial conflict relevant to a specific publication should be disclosed clearly.

Material errors should be corrected promptly. A material correction should identify what changed, preserve the distinction between the previous and corrected interpretation, and rebuild affected outputs when practical.

Trust should come from transparent definitions, honest limitations, repeatable validation, and visible correction, not from credentials or independence alone.

Contact

Vyreon Labs is based in Calgary, Alberta, Canada.

Feedback is useful. If you read the reports, use the charts, or have questions about the methodology, contact Vyreon Labs through LinkedIn.

Contact on LinkedIn

For employment, technical collaboration, methodology questions, corrections, or data-access inquiries:

Sheldon Glowatski
Mechanical Engineer and APEGA Engineer-in-Training
Email: sheldon.glowatski@ucalgary.ca
LinkedIn: Sheldon Glowatski

Author: Sheldon Glowatski
Professional registration: Engineer-in-Training, APEGA
Location: Calgary, Alberta, Canada
Last reviewed: August 2026