PeakProcess Industrial Consulting ยท by Rok Erman, PhD

Problem on the production floor? Let us fix it together.

Industrial diagnostics, process optimization, and AI for production lines where problems persist, quality drifts, or throughput falls short of target.

See showcases Request a conversation
15+ yearsindustrial analytics and AI
25+ projectsmanufacturing, pharma, finance, gaming
Low-friction deliveryworks without a big data team

The problem

When production problems refuse to go away

Many industrial organizations collect data, run tests, and adjust parameters โ€” but still struggle to find the real cause of recurring defects, unstable processes, or quality problems. The issue is rarely lack of effort. It's lack of the right diagnostic model.

๐Ÿ”
Recurring defects with no clear root cause

The problem keeps coming back despite repeated fixes. The underlying cause has not been correctly identified.

๐Ÿ“‰
High scrap or rework costs

Yield losses that are accepted as normal but represent significant recoverable margin.

โš ๏ธ
Unstable production parameters

Process behaviour varies across shifts, lines, or batches with no systematic explanation.

๐Ÿ‘๏ธ
Quality-control systems that miss defects

Inspection passes parts that fail downstream, or rejects too many good parts โ€” either way, the system is not calibrated correctly.

๐Ÿšง
Bottlenecks reducing throughput

One step limits the whole line and the constraint has not been correctly located or quantified.

๐ŸŽฒ
Too much trial-and-error decision-making

Experiments are run without a structured hypothesis. Results are inconclusive or misleading.

๐Ÿ’พ
Production data collected but not used

Sensor data, MES logs, and QC records exist but have never been analysed in a way that produces actionable insight.

๐Ÿงฑ
Internal teams stuck without a diagnostic model

Skilled engineers are working hard but without a mathematical or AI framework to guide the diagnosis โ€” so the effort doesn't converge.

Use cases

The proof!

Impact delivered
+3% Throughput
Problem: recurring specification failure on a production line

Consumer Appliances / Manufacturing

Coffee grinder production line stabilization

A persistent quality issue caused measurable throughput loss. Structured diagnosis and targeted recalibration delivered a verified +3% throughput gain.

The situation: A high-volume appliance line had a recurring grinder specification failure with no clear root cause.

What I did: Sequenced hypotheses by testability, isolated key parameters, and corrected a shift-sensitive calibration drift.

Technical approach: SPC analysis, measurement-system validation, and targeted parameter recalibration.

Result: Verified +3% throughput with stable post-fix performance.

Impact delivered
Solved in weeks
Problem: multi-year injection-molding defect (rotor cracking)

Automotive / Injection Molding

Fuel pump rotor cracking elimination

A defect unresolved for years across teams in two countries was resolved in weeks using structured experimentation across interacting process parameters.

The situation: Persistent cracking created supply-chain and delivery risk.

What I did: Built and executed a DoE matrix over temperature, pressure, hold, and material composition interactions.

Technical approach: Process-window mapping + material behaviour analysis.

Result: Defect removed within a stable operating window in weeks.

Impact delivered
Sub-4 ฮผm Precision
Problem: multi-sensor QC alignment for production-speed inspection

Automotive / Precision Metrology

End-of-line quality-control calibration

Designed calibration logic across camera and ToF systems to meet extreme precision constraints at production speed.

The situation: Three measurement systems in separate coordinate spaces had to operate as one.

What I did: Built a robust transformation/alignment pipeline with uncertainty validation.

Technical approach: Multi-sensor calibration with certified reference geometry checks.

Result: Sub-4 ฮผm precision enabled production-speed QC.

Pharma / Regulated Manufacturing

Generic drug development system

Estimated 20% faster time-to-market through end-to-end process management and decision support.

Problem: Fragmented development workflow and slow decision cycles.

Impact: Estimated 20% time-to-market reduction with sustained internal ownership.

Pharma / Process Optimization

Reverse osmosis filtration productivity

Measured +0.2% productivity increase in a constrained regulated environment.

Problem: Throughput bottleneck in filtration step with limited safe operating range.

Impact: Measured +0.2% productivity at industrial scale.

Industrial Equipment / Root Cause Analysis

Production fault attribution framework

Built a practical framework classifying faults into human, equipment, and material causes for faster corrective action.

Problem: Repeated failures with no consistent root-cause taxonomy.

Impact: Faster diagnosis and better prioritization of corrective actions.

How this works

Senior expert. No handoffs. No junior layer.

Clients work directly with Rok. The person diagnosing the problem is the person doing the work โ€” not a project manager relaying information to a team you never meet. No generic agency process, no account-manager filter, no recycled consulting templates.

PhD-level mathematical and statistical reasoning
15+ years of cross-sector industrial and analytical experience
Deep hands-on knowledge of manufacturing, pharma, and industrial systems
Practical focus โ€” measurable outcomes, not methodology reports
Strong fit for unclear, technically complex, commercially expensive problems

Entry point

Diagnostic Sprint

A focused 1โ€“2 week technical diagnostic for companies with recurring production defects, unstable processes, or quality-control problems that internal teams have not been able to resolve.

Review of available production data and process parameters
Failure pattern analysis across shifts, batches, or lines
Root-cause hypotheses ranked by probability and testability
Mathematical or AI model where it adds real diagnostic value
Clear written findings and prioritised action plan
Recommendation for next step: prototype, implementation, or follow-on project
Start with a Diagnostic Sprint

Engagement options

Three ways to work together

๐Ÿ” Diagnostic Sprint

1โ€“2 weeks. Focused root-cause diagnosis with a clear findings report and action plan. Best starting point for a new or unsolved problem.

โš™๏ธ Project-based implementation

Scoped engagement to build, validate, and hand over a model, system, or AI use case. Includes training for the internal team to operate it independently.

๐Ÿ“… Monthly advisory retainer

Ongoing access for organizations that want a senior technical partner available to review decisions, unblock problems, and guide the team as work progresses.

How most engagements start

"Send a short description of the issue, what has already been tried, and what data is available. I'll assess whether a diagnostic sprint is a good fit โ€” and if it isn't, I'll say so."

Most clients start with a Diagnostic Sprint. The findings either close the problem or define a clear scope for implementation.

How I work

A clear path from first diagnosis to full-scale execution

1
Always included

Diagnostic

Map the process, identify the actual pain point (not the assumed one), assess data reality, and quantify the economic impact of solving it. This step alone often surfaces value.

2
Always included

Solution design

Select the right technical approach โ€” digital twin, ML model, optimization algorithm, computer vision, or a combination. Design the implementation path and define what success looks like before any code is written.

3
Always included

Pilot & handover

Deliver a working pilot validated against real production data. Train the internal team to operate it. Hand over documentation, code, and a clear roadmap for what comes next โ€” so the organization owns the outcome.

Optional extensions

๐Ÿ—‚ Optional add-on

Execution coordination

When the roadmap is complete, I can stay on to coordinate the full rollout โ€” managing the implementation sequence, resolving technical blockers, and keeping the programme on track from pilot to production-scale deployment. One consistent point of accountability throughout.

๐Ÿ‘ฅ Optional add-on

Team assembly & leadership

For organizations that don't have the internal capability to execute, I can assemble and lead the right team โ€” sourcing data engineers, ML engineers, or domain specialists as needed, and directing the work from scoping through delivery. You get a functioning team without the overhead of hiring one yourself.

About

Rok Erman PhD

Rok Erman is a PhD mathematician with 15+ years of experience applying mathematical modelling, AI, analytics, and structured problem-solving across manufacturing, automotive, pharma, finance, and gaming.

His strongest work is in situations where the problem is technically complex, commercially expensive, and not solved by standard trial-and-error. He brings a rigorous diagnostic approach that identifies what is actually driving the problem โ€” then turns that into practical recommendations, working models, or a clear implementation roadmap.

He works directly with clients โ€” no junior intermediaries, no agency layer. The engagement is as light or as deep as the problem requires.

EducationPhD in Mathematics
Experience15+ years
SectorsManufacturing, Automotive, Pharma, Finance, Gaming
MethodsMathematical modelling, ML, computer vision, DoE, process optimization
LanguagesSlovenian, English, German
LocationSlovenia โ€” works internationally

Get in touch

Have a production problem your team cannot clearly explain?

Send a short description of the issue, what has already been tried, and what data is available. Rok will assess whether a diagnostic sprint is a good fit.

Diagnostic Sprint1โ€“2 week focused diagnosis with a findings report and action plan.
Intro call30 minutes to assess fit and scope the problem.
Direct emailrok.erman@gmail.com

After submit, your message is emailed to rok.erman@gmail.com.