Philipp Bühler

Student · Entrepreneur

I design and builddecision systems —software, models,businesses.

I am a master’s student and I run Bühler Digital Solutions. Most of my time goes into web platforms and AI-assisted workflows for real businesses. The rest goes into a thesis asking how a reinforcement-learning agent holds up against classical optimization when the goal is risk-adjusted return.

Selected workResearchContact
Diagram of one path through five connected domainsFive stacked horizontal layers labelled Engineering, AI & Data, Product & Business, Entrepreneurship, Research & Finance, threaded by a single continuous line that passes through one node on each layer.ENGINEERINGAI & DATAPRODUCT & BUSINESSENTREPRENEURSHIPRESEARCH & FINANCE
One path, not five separate skills
Profile

Short version

I work where technology, business and analytical decision-making meet. What ties my projects together is an interest in turning complex or uncertain problems into systems that are structured, understandable and useful in practice.

I start with the structure underneath a problem. The assumptions it rests on, the decisions it hides, and the complexity it does not actually need. Clarity matters, though not at the cost of what is essential.

Technology should extend what people can do without taking away their understanding, judgment and responsibility. The systems I trust most make their own limits visible and stay open to being questioned, adapted and improved.

“The best way to predict the future is to invent it.”

Alan Kay

A boundary worth stating

This is a personal site. Bühler Digital Solutions is a separate company with its own website. It is named here because it is part of the work, not because this page is selling anything.

Portrait to follow
Based in
Germany, Baden-Württemberg
Currently
Master’s studies · entrepreneurship
Working on
Web platforms · AI workflows · portfolio optimization · trading bot
Background
Mercedes-Benz · KPMG · self-employed
Selected work

What I have built

Three things worth showing. The practice I run, the thesis I am writing, and the first site of mine that went live.

Operating

Bühler Digital Solutions

Company

Context
Most small businesses need one digital thing to work properly. A site that earns its keep, or a process that stops eating an afternoon every week. What they usually get offered instead is an agency retainer.
Approach
A one-person practice, scoped per project rather than sold as a package. Web development and design, digitalisation, AI-assisted solutions, automation of business processes, search visibility, and maintenance where it has actually been agreed.
Role
Founder

Stack

  • Web development
  • Web design
  • AI solutions
  • Automation
  • SEO
  • Maintenance

Company website URL

In progress, no results yet

DRL Portfolio Optimization

Master’s thesis · research

Context
Mean-variance optimization solves a static problem at every rebalancing date and leans hard on an estimated covariance structure. It is elegant, and it is the benchmark for good reasons. It also assumes a stability that markets do not offer.
Approach
A PPO agent is trained to allocate across assets under three risk profiles, then measured against mean-variance optimization and a market portfolio on the same data under the same constraints. Explainable-AI methods are scoped as an optional extension, so the agent’s allocations can be questioned rather than only scored.
Role
Author, thesis in progress

Stack

  • Python
  • PPO / reinforcement learning
  • Portfolio theory
  • Backtesting
Live

Gesundheitspraxis Johe

Website · first project to go live

Context
A health practice with no web presence at all. It was the first site of mine to go live. I built it as a favour rather than as a paid engagement, so it belongs here as a reference and not as a case study.
Approach
A static, fast site on the stack I use for client work, with a content layer the practice can maintain without me. The full write-up covering brief, constraints and outcome is not published, and nothing is claimed here that has not been confirmed.
Role
Design and implementation

Stack

  • Astro
  • React
  • Tailwind CSS
  • Keystatic CMS
  • Cloudflare Pages

Visit (opens in a new tab)

Research · master’s thesis

Reinforcement learning for portfolio allocation

Whether a learned policy can beat a benchmark on one backtest is not a very interesting question. That result is cheap and rarely survives. What I want to know is whether a policy trained under an explicit risk profile allocates differently from a classical optimizer, and whether the difference still holds when both are handed the same data, the same universe and the same constraints.

Schematic risk–return comparison, not measured resultsA risk versus return chart showing the comparison being set up. A curve marks the mean-variance efficient frontier and a single point marks the market portfolio. At each of the three risk profiles (Conservative, Balanced, Aggressive) the PPO agent is drawn as a vertical bracket with a question mark, because its return is not yet known. No result is shown.CONSERVATIVEBALANCEDAGGRESSIVEMVOMARKET???RISK →RETURN →
  • Mean-variance frontier
  • Market portfolio
  • PPO, not yet measured
Schematic, not measured. The axes show how the comparison is set up, with risk against return and three strategies across three risk profiles. The positions are illustrative and claim nothing about outcomes.

Working title

Deep-reinforcement-learning-based portfolio optimization using PPO, compared against mean-variance optimization and a market portfolio.

Strategies compared

  • PPOLearned allocation policy, trained per risk profile
  • Mean-VarianceThe classical optimizer, and the benchmark
  • Market portfolioPassive reference

Scope

  • Proximal Policy Optimization (PPO)
  • Mean-variance optimization (MVO)
  • Market portfolio as a passive reference
  • Three risk profiles
  • Risk / return trade-off
  • Explainable AI (optional extension)
Experience

Where the work happened

Roles and periods come from me. Where tasks and outcomes are not published, it says so instead of being filled in.

  1. 2026–presentCurrent

    Bühler Digital Solutions

    Founder

    Web development and design, digitalisation, AI-assisted solutions, automation, search visibility, and maintenance for small and local businesses.

  2. 2026–2027Current

    Master’s thesis · research

    Author, deep reinforcement learning for portfolio optimization

    PPO-based portfolio allocation across three risk profiles, benchmarked against mean-variance optimization and a market portfolio. Still in progress, with no results yet.

  3. 2024–2027Current

    Reutlingen University

    M.Sc. Business Informatics (Wirtschaftsinformatik)

    The master’s programme is ongoing, and the thesis above is part of it.

  4. 03/2025–03/2026

    KPMG

    Working Student

    Tasks, focus and outcomes are not published here.

  5. 2022–2023

    Mercedes-Benz AG

    Working Student · Intern

    Tasks, focus and outcomes are not published here.

  6. 2020–2024

    Reutlingen University

    B.Sc. Business Informatics (Wirtschaftsinformatik)

    The undergraduate degree that came before the master’s programme.

Capabilities

Four kinds of problem

Grouped by the kind of problem rather than by tool. These are areas I actually work in, not a certificate list and not a wall of logos.

Engineering

Building the thing so it survives contact with a second developer.

  • Java · JavaScript · TypeScript
  • HTML & CSS
  • Astro & static site generation
  • Component architecture
  • Database design & management (SQL)
  • Accessibility and performance budgets
  • Git-based workflows

AI & Data

Getting useful, checkable work out of models and data.

  • Machine learning & data science (Python)
  • Reinforcement learning (PPO)
  • LLM-assisted engineering workflows
  • Agent orchestration & separated responsibilities
  • BI dashboards (Power BI, Tableau, SAP Analytics Cloud)

Product & Business

Deciding what is worth building, and for whom.

  • Project management
  • Scoping and requirement clarification
  • Business process automation
  • Consulting
  • Search visibility (SEO)
  • Teamwork & communication

Research & Finance

The quantitative side, and the honesty it demands.

  • Academic research & scientific writing
  • Portfolio theory
  • Mean-variance optimization
  • Risk / return analysis
  • Backtesting methodology
  • Explainable AI
Development

Build activity

Public GitHub activity, read when the site is built. Only what the API actually returns is shown, with no invented figures and no numbers padded to look busier.

Public repositories

Languages in use

  • Python
    96.2%
  • PowerShell
    3.8%

Profile

@
Belaunsch
On GitHub since

View profile on GitHub (opens in a new tab)

Data fetched ·
Refreshed each time the site is rebuilt.

Current focus

Open right now

What is actually open on the desk right now.

  • Ongoing

    Building Bühler Digital Solutions

    Turning a practice into something repeatable. Offer structure, delivery standards, and the maintenance side that most one-person operations under-plan.

  • In progress

    Master’s thesis on DRL and portfolio optimization

    Setting up the PPO agent and the risk profiles, and building a comparison against mean-variance optimization and a market portfolio that is fair by construction.

  • Continuous

    AI-assisted engineering workflows

    Working out where an agent genuinely reduces effort, and where it only moves the review burden somewhere less visible.

Contact

Get in touch

LinkedIn is the fastest route to me. The email address is deliberately not printed here yet.