Voron 2.4 — ABS-Capable CoreXY
I spent summer 2025 resurrecting a Voron 2.4 kit that arrived with bent frame members and a flaky CAN toolhead. The process covered mechanical realignment, firmware-level debugging, and heat management for high-temperature materials.
Key Fixes
- Straightened extrusions, re-squared gantry, and dialed in belt tension to remove layer shifts.
- Re-terminated CAN cabling and reflashed toolhead boards to eliminate intermittent communication faults.
- Upgraded hotend cooling and enclosure sealing so the printer is dependable with ABS and ASA.
Impact
- Weekly ABS prints without downtime thanks to preventative maintenance routines.
- Deep familiarity with Klipper macros, CAN bus diagnostics, and Voron serviceability.
- A platform ready for future experiments such as toolhead swaps and pellet extrusion tests.
High-Density Metal 3D Printing Thesis
My bachelor's thesis at Bosch focused on inventing a filament-based metal additive process that could stand next to long-established MIM lines. I owned material formulation, printer process setup, and the statistical modeling that guided us toward the winning parameter windows.
What I Built
- Developed multiple metal-polymer filament recipes and characterised their flow, shrinkage, and sintered properties.
- Ran a structured Design of Experiments campaign to map pressure-sintering variables to final density and mechanical performance.
- Pioneered the "Bricklayer" deposition strategy—alternating line offsets between layers—to eliminate voids and lift green-part density before sintering.
Key Results
- Identified two optimal print + sinter settings that delivered densities approaching benchmark MIM samples.
- Validated Bricklayer printing with micrographs that showed reduced porosity versus conventional toolpaths.
- Optimised print temperatures and flow rates for both proprietary and off-the-shelf filaments, enabling repeatable builds on Bosch hardware.
What I Learned
- How to design DOE matrices, run them inside a corporate R&D lab, and translate the stats into production guidance.
- How to secure lab time and cross-team resources in a large organisation by communicating results and next steps clearly.
- Why material science rigour—sample prep, microscopy, traceability—matters as much as the printer tuning itself.
Automated Note Taking Pipeline
The system starts with raw audio, pushes transcripts through OneDrive, and orchestrates prompt templates that condense hours of lectures into concise summaries. Early versions used fragile UI automation; later iterations migrated to Gemini and API-driven steps for durability.
Workflow Highlights
- Local watcher script triggers Whisper V2 transcription for any new recording.
- Power Automate handled the initial Claude integration before being replaced by API-based connectors.
- Output packaged as markdown and spaced-repetition flashcards for quick review.
Lessons
- UI automation is brittle; wherever possible convert to API calls for stability.
- High-context LLMs collapse multi-step prompts into lighter, faster pipelines.
- Reliable logging and retry logic are just as important as prompt quality.
Plexiglass Light Sign Gift
This project combined machining plastics with simple electronics. I dialed in feeds and speeds to avoid melted edges, then wired a diffused LED strip into a wooden housing that makes the engraved artwork glow.
Build Notes
- Experimented with shallow passes and chip clearing to keep the acrylic cool during engraving.
- Designed a plywood base with channels for LEDs and diffusion film for even lighting.
- Finished the stand with Danish oil and discreet cabling so the gift looked polished.
Takeaways
- Plastics demand different machining heuristics than wood or aluminum.
- Lighting design is as important as the engraved art for the final effect.
- Small, personal builds sharpen attention to aesthetic details and finishing.
Efficient Company Concept
The concept distills lessons from hardware builds, AI pipelines, and consulting sprints. Every successful engagement shared the same DNA: document decisions, automate repetitive work immediately, and keep teams tiny but multidisciplinary.
Highlights
- Defines automation-first workflows, radical transparency, and KPI discipline as core principles.
- Maps an operating stack spanning knowledge bases, modular prompts, automation hubs, and customer dashboards.
- Captures experiments already run across Voron rebuilds, LeanGPT, and consulting projects to prove the framework.
Why It Matters
- Gives structure to future venture pilots so they launch with measurable traction.
- Clarifies hiring profiles and tooling decisions for a lean, high-leverage team.
- Builds a reusable SOP library that compounds with every project.
DIY CNC Mill Build
I designed and assembled a Mostly Printed CNC (MPCNC) machine from scratch to gain hands-on experience with digital fabrication. The project forced me to balance budget constraints with performance, source components, print structural parts, and ultimately calibrate a precise motion system.
Highlights
- Sourced and integrated stepper motors, controller boards, and wiring, learning to diagnose issues such as reversed motor polarity.
- Replaced a failed control board, reinforcing a habit of structured debugging and careful electronics handling.
- Tuned the machine using Estlcam and Fusion 360 post-processors so it could execute complex 3D toolpaths reliably.
Impact
- Produced 50+ engraved and milled gifts, demonstrating repeatable accuracy.
- Developed a repeatable workflow from CAD design to G-code, which I now apply to other automation projects.
- Documented calibration lessons that continue to inform my CNC-to-robot conversions.
3D Printing Mastery
3D printing was my entry point into engineering. I moved beyond downloaded models by adopting Fusion 360, iterating on custom fixtures, and stress-testing different materials such as PLA, PETG, and ABS.
What I Learned
- Design for additive manufacturing: factoring in wall thickness, support strategy, and tolerances during CAD work.
- Material science basics: selecting filaments based on strength, heat resistance, or flexibility requirements.
- Process optimization: tuning slicer profiles for reliability, leading to the 95% print success metric.
Why It Matters
- Rapid prototyping accelerates proof-of-concept cycles for everything from robotics jigs to custom electronics enclosures.
- Hands-on experience with manufacturing constraints directly improves my mechanical and product design decisions.
- Community engagement through forums and maker spaces keeps me current on best practices and upcoming technologies.
Python Automation Suite
I automated repetitive gearbox calculations required for my mechanical engineering coursework. The desktop app performs data validation, orchestrates the formulas, and presents results in a clean UI, removing human error from the workflow.
Technical Stack
- Python with Tkinter for the GUI layer and structured input handling.
- Modular calculation engine to encapsulate formulas and allow future extensions.
- Automated QA scripts to compare calculator output with verified manual solutions.
Results
- Saved at least 58 minutes per assignment while improving consistency across student teams.
- Demonstrated the business value of lightweight automation to faculty and peers.
- Continues to serve as a foundation for newer engineering toolkits I am developing.
The source code is published on GitHub: GertiebeD.
AI Integration Projects
My AI experiments focus on practical productivity gains. I built an education assistant that auto-generates flashcards, prototyped a lecture transcription pipeline, and used Stable Diffusion to craft consistent digital characters.
Key Builds
- LeanGPT: A Python + Flet application for summarising lectures and generating study aids.
- Automated note-taking system: Whisper V2 for transcription, Microsoft Power Automate for workflow orchestration, and Claude for high-token analysis.
- Generative art workflows: Custom prompts and parameter tuning in Stable Diffusion to prototype realistic personas.
What I Took Away
- API reliability, latency, and cost control are as important as model accuracy.
- Prompt engineering is an iterative design exercise that mirrors UX research.
- Even partial automations deliver value when paired with human oversight.
Ungetüm E-Bike
I wanted a vehicle that balanced stability and carrying capacity, so I built a bespoke trike. The design combined reclaimed bicycles, a welded frame, and an electric drive system powered by repurposed car batteries.
Engineering Focus
- Motor controller tuning for smooth acceleration without overheating.
- Weight distribution analysis to maintain grip and safe braking distances.
- Iterative steering geometry adjustments to avoid wobble at 18 km/h top speed.
Lessons Learned
- Brake systems must be sized for total vehicle mass, not just component specs.
- Battery management is critical when combining cells with different histories.
- Rapid prototyping with wood and steel speeds validation before committing to final welds.
Systems Thinking Framework
Understanding how components interact is at the core of my problem-solving style. I map systems by defining their purpose, elements, relationships, and external influences so I can uncover leverage points and failure modes.
Applications
- Diagnosing performance bottlenecks in multi-step manufacturing workflows.
- Planning renovation tasks by sequencing dependencies and risk mitigation.
- Translating social or business systems into actionable models for stakeholders.
Why It Matters
- Systems thinking enables reliable decision-making even with incomplete data.
- It helps me communicate technical trade-offs clearly to diverse audiences.
- The framework scales from hardware builds to organisational processes.
Creative Writing Projects
Writing is where I test narrative structure and long-term project discipline. "Der Krieger" follows a hero’s evolution across three planned books, while "The Spark" experiments with parallel storytelling to show how small changes can reshape systems.
Creative Skills
- World-building and character arcs rooted in systems behaviour.
- Long-form editing cycles that mirror iterative product development.
- Balancing symbolism with accessible storytelling for broader audiences.
Solo Room Renovation
With limited resources and no on-site help, I renovated the top floor of my parents' house to gain privacy and a focused workspace. The project demanded planning, trade coordination, and strict safety practices.
Scope of Work
- Installed insulation, drywall, lighting, and outlets to modern standards.
- Repaired flooring and integrated new windows to improve natural light.
- Designed and built space-saving furniture tailored to the room geometry.
Takeaways
- Large projects succeed when broken into disciplined, manageable stages.
- Documentation and measurement prevent rework, especially when working solo.
- Resourcefulness—repurposing materials, adapting tools—kept the budget in check.
Market & Sales Strategy for a Pharmaceutical Lab
As project lead on a university-industry collaboration, I coordinated a four-person team helping a pharmaceutical testing laboratory expand beyond its home country. Field research in Germany confirmed the lab was invisible to overseas buyers, so we reframed the entire sales strategy around how conservative labs actually build trust.
My Role & Key Contributions
- Ran the market discovery work, including in-country SEO checks that revealed no international search traction.
- Devised a direct outreach strategy centred on cold-calling lab directors and showing up at targeted conferences.
- Facilitated team synthesis sessions and owned all customer and competitor analysis deliverables.
Challenges & Solutions
- When enterprise buyers ignored surveys, we pivoted to deep qualitative interviews that uncovered a niche of smaller labs without certified testing gear.
- Translated sensitive interview insights into an actionable playbook that respected industry expectations for relationships and compliance.
Skills Demonstrated
- Market analysis across SWOT and PESTLE lenses.
- SEO diagnostics and competitive positioning.
- Go-to-market strategy and team leadership.
Advanced Composite Material for Orthopedic Insoles
Partnering with a materials science company, I led a four-week R&D sprint to stop structural creep in custom orthopedic insoles. We combined a flexible FDM-printed shell with secondary fillers that deliver higher compressive strength while preserving the artisan feel shoemakers expect.
My Role & Key Contributions
- Developed the composite concept by experimenting with synthetic rubber, silicone, and PU foam infill options.
- Conducted interviews with traditional shoemakers to align product requirements with how they fit and finish insoles.
- Built a SaaS-centric business model where shoemakers subscribe to design software and upsell proprietary filler material.
Challenges & Solutions
- Initial prototypes underperformed pure TPU, so we benchmarked filler materials and recommended a higher-strength core for phase two.
- Balanced manufacturability with post-processing flexibility so craftspeople could make last-minute adjustments.
Skills Demonstrated
- Material science experimentation and 3D printing process control.
- User-centred design and qualitative interviewing.
- Business strategy and SaaS monetisation models.
Waterproof Housing for a Wearable Heart Rate Sensor
For my Product Development and Manufacturing course, I owned the enclosure design for a wearable heart rate monitor. The goal: build a design that could start life as a low-volume prototype but transition seamlessly to mass production without rework.
Design & Engineering Highlights
- Modelled all parts in Fusion 360 with proper draft angles, parting lines, and wall thicknesses for injection molding.
- Delivered a waterproof architecture using an O-ring seal and translucent housing that allowed hidden LEDs to shine through.
- Integrated a capacitive touch button that remained responsive through the sealed shell.
Skills Demonstrated
- Mechanical design and design-for-manufacture principles.
- Injection molding readiness and tolerance planning.
- Waterproof enclosure engineering.
Sustainable Supply Chain for a Window Manufacturer
Leading a consulting team for a Spanish window manufacturer, I mapped a new supply chain that replaced petroleum-based polymers with sustainable alternatives. The work spanned lab validation, partner scouting, and a circular end-of-life plan.
My Role & Key Contributions
- Discovered the core academic research enabling the material swap and translated it into an operational roadmap.
- Identified European collection partners and processors, cutting logistics lead time by 1.5 months.
- Designed a closed-loop system that recycles end-of-life frames back into new foam production.
Skills Demonstrated
- Supply chain design and logistics optimisation.
- Circular economy and sustainability analysis.
- Project management across technical and business stakeholders.
Go-to-Market Strategy for a Cryogenic Technology Firm
In a high-tech consulting project, my team developed the market expansion plan for a precision cryogenic system. I focused on how the company could build trust quickly in an emerging quantum computing ecosystem.
My Role & Key Contributions
- Architected an ingredient-branding approach by partnering with established cryogenic station manufacturers.
- Persuaded the CEO to adopt authentic educational content over polished advertising, using case studies from analogous industries.
- Mapped a content calendar highlighting technical deep dives and video explainers hosted on the platforms scientists already use.
Skills Demonstrated
- Market entry strategy and partner development.
- Stakeholder management at the executive level.
- Brand positioning and competitive analysis.
Data Science Study: Elon Musk's Tweets vs. Tesla Stock
This team project tested the assumption that social media activity from a CEO could meaningfully sway stock prices. We analysed over a decade of tweets and market data, validating model performance with human-in-the-loop reviews.
Methodology & Findings
- Constructed a Python data pipeline combining Twitter API exports with Tesla stock data inside Pandas.
- Used a Llama 3 language model for sentiment analysis and spot-checked results through a HITL workflow to handle irony and emojis.
- Ran linear regression models that showed no statistically significant correlation at the daily level, indicating broader market forces dominate.
Skills Demonstrated
- Data science tooling in Python, including Pandas and regression analysis.
- NLP with large language models and quality assurance via HITL.
- API integration and collaborative research practices.
Two-Stroke Engine Prototype
Over 2 months during my school years, I built a complete two-stroke engine targeting 15 HP output. While the engine never sustained operation due to crankshaft bearing issues, the project delivered invaluable lessons in precision machining, engine fundamentals, and failure analysis.
Technical Approach
- Designed complete engine assembly in CAD including cylinder, piston, crankshaft, and housing
- Machined all components in school workshop using manual and CNC equipment
- Achieved excellent cylinder-to-piston tolerances without proper piston rings—demonstrating precision machining fundamentals
The Fatal Flaw
- Crankshaft ran with excessive friction, causing it to stick under combustion pressure
- Welded joints in the crank arm repeatedly failed under concentrated forces
- Engine destroyed itself during testing rather than achieving sustained operation
Key Learnings
- Two-stroke theory: port timing, air-fuel mixture dynamics, and combustion cycles
- Bearing design: The critical importance of low-friction rotating assemblies
- Failure analysis: Understanding why something fails teaches more than building something that works
- Manufacturing precision: Sub-millimeter tolerances on manual equipment
CNC Box Production
Using my DIY CNC mill, I manufactured custom engraved boxes from laminated wood. The project became a masterclass in batch manufacturing: the first box took hours, the last took a quarter of that time. Setup and tooling dominated the process more than raw machine speed.
The Evolution
- Boxes 1-3: Manual alignment, frequent re-zeroing, 3-4 hours per box
- Boxes 4-8: Built alignment jigs and quick-release clamps, 2 hours per box
- Boxes 9-20: Refined fixtures enabled drop-in positioning, 45-60 minutes per box
Manufacturing Principles Discovered
- Setup time dominates small-batch production (under ~50 units)
- Fixtures pay for themselves: even simple jigs deliver massive time savings
- Repeatability enables speed: proven processes run confidently without constant verification
- Process design matters as much as machine capability
Business Insight
- Time invested in tooling is quickly recovered in production runs
- Standardization enables scaling: once the process works, replication is cheap
- Small-batch manufacturing is dominated by setup overhead
Wood Hardening Experiments
Inspired by academic research into chemical wood modification, I experimented with processes that fundamentally transform wood's material properties. The treated wood became significantly harder and denser, opening possibilities for tool applications while teaching material science fundamentals.
Experimental Results
- Softwood: ~50% volume reduction, significantly increased density and hardness
- Hardwood: ~30% volume reduction, noticeable hardness improvement
- Surface properties: Metal tools required substantially more pressure to scratch the surface
- Structural integrity: Material retained flexibility without becoming brittle
Application Testing: Wooden Knives
- Successfully shaped cutting edges from hardened wood
- Could cut vegetables and soft foods initially
- Edge dulled after first cut—impractical for repeated use
- Validated hardness transformation while revealing practical limitations
Practical Applications
- Tool handles: increased durability for grips and handles
- Wear surfaces: harder material suitable for jigs, fixtures, and bearing surfaces
- Decorative elements: altered appearance and feel for aesthetic uses
CNC Painting Robot
I transformed my CNC mill into an automated artist by designing a tool-changing system for pens and brushes. The real challenge wasn't mechanics—it was generating G-code for art. By hacking Prusa Slicer's multi-material workflow, I created a pipeline from digital artwork to physical output that was orders of magnitude faster than traditional CAM software.
Technical Innovation
- Tool system: Manual tool-changing with self-aligning, gravity-fed pen/brush holders
- 48-color palette: Works with both colored pencils and acrylic paint
- G-code generation: Repurposed Prusa Slicer by treating art layers as single-height 3D prints
- Multi-color orchestration: Slicer automatically generates tool-change pauses between colors
The Software Breakthrough
- Standard CNC software was painfully slow for artistic paths
- Discovered Prusa Slicer could color-separate SVG files and coordinate tool changes
- Modified slicer profiles to disable 3D movement, optimize for 2D drawing
- Achieved faster workflow than purpose-built CAM tools
Results & Impact
- Created wall art with sub-millimeter precision impossible to achieve manually
- Pencil artwork highly reliable; acrylic paint more challenging but painterly
- Experimented with multi-layer color mixing for shading effects beyond base palette
- Artwork decorated walls before relocating for master's program