Freelance projects · open to permanent roles in Germany

Product · AI · Backend · Infrastructure

I build software products, AI systems and the infrastructure behind them.

I work across product decisions, application architecture, data, local AI, Linux and delivery—using modern AI engineering tools to move faster, then verifying the result with tests, CI and real-system evidence.

Applied AI systemsFull-stack productsPostgreSQL & APIsLinux & infrastructureCI/CD & release

Selected systems

Proof across different kinds of engineering.

Four projects show the range better than a long technology list: applied AI, local inference, consumer product architecture and transactional business systems.

01Applied AI · Developer tools

Diagnostic reasoning platform

Rikaha

A diagnostic reasoning system for software troubleshooting, with a separate Intelligence Lab for benchmarks, evaluators and controlled engine evolution.

  • Progressive diagnostic state instead of one-shot scenario matching
  • Separate product/runtime and research/engine authority
  • Protected benchmark, evaluator and provenance discipline
TypeScriptFastifyNext.jsPostgreSQLApplied AI
Active platform + research
Case study →Private source
02Local AI · Systems engineering

Private persistent AI system

ROS

A local-first assistant architecture that separates durable identity and state from replaceable models, with secure GPU inference, bounded capabilities and operational recovery.

  • PostgreSQL-backed persistent assistant state
  • Dedicated GPU inference worker over TLS 1.3 mTLS
  • Model benchmarking, role policy and no-cloud-fallback boundaries
PythonPostgreSQLLinuxLocal LLMsmTLS
Private working system
Case study →Private source
03Consumer product · Recommendation intelligence

Discovery + community platform

DramasPicks

A drama discovery product that keeps catalogue truth, human opinion and derived recommendation intelligence separate while supporting community, profiles and operational workflows.

  • Canonical catalogue and provider provenance
  • Community, privacy and moderation-aware domain modelling
  • Separate Intelligence Lab for recommendation evidence and evaluation
ReactTypeScriptNode.jsPostgreSQLPlaywright
Active product
Case study →Private source
04Commerce · Transactional systems

Farm-to-consumer platform

AndhraFresh

A farmer-first commerce system built around real physical supply, explainable pricing, safe order commitment, fulfillment and settlement back to the farmer.

  • Traceable farm → harvest → supply → order relationships
  • Concurrency-safe allocation and real PostgreSQL contention tests
  • Pricing, payment state, fulfillment, settlement and payout foundations
TypeScriptPostgreSQLAPIsDomain modellingCI/CD
Substantial working foundation
Case study →Private source

Engineering range

Useful when the problem crosses boundaries.

My strongest work is where product, software, AI and operations meet—not where each layer is treated as somebody else's problem.

01

Applied AI systems

Local inference, persistent AI state, model evaluation, reasoning engines, secure workers and research-to-product boundaries.

Local LLMsEvaluationInferenceAI product architecture
02

Product engineering

Consumer and business products that connect UI, APIs, domain rules, data, operations and release workflows.

React / Next.jsTypeScriptAPIsProduct systems
03

Backend & data

PostgreSQL-backed systems with migrations, provenance, transactional state, concurrency, audit and recovery.

PostgreSQLTransactionsDomain modellingData lineage
04

Platform & delivery

Linux environments, Docker, self-hosted CI, networking, observability, packaging and controlled deployment.

LinuxDockerGitHub ActionsCloudflare

More systems

The engineering behind the products.

Supporting work covers editorial workflows, infrastructure truth, Linux operations, observability and reusable developer tooling.

Workflow product

BriefSift

Source-linked editorial workflow with public reader, private newsroom, ingestion, publish guards and PostgreSQL migrations.

View case study →

Platform / operations

Rikaha Labs Infrastructure

Linux, Proxmox, OPNsense, DNS, self-hosted CI, recovery paths and controlled service operations.

View case study →

Systems tooling

Control Station

Provenance-aware infrastructure truth: immutable observations, freshness, conflicts and read-only collectors.

View case study →

Developer tooling

Node Platform

Versioned Linux operator tooling with Debian packaging, signed APT delivery, canary rollout and rollback.

View case study →

AI workflow / developer tools

Engineering OS

Deterministic Python authority, lifecycle and readiness tooling for recoverable AI-assisted engineering.

View case study →

How I work

AI-assisted engineering, with ownership kept human.

ChatGPT and Codex are part of my day-to-day implementation, research and debugging workflow. I use them as force multipliers—not as a substitute for defining the problem, making architecture decisions or proving the system works.

  1. 01

    Frame

    Define the product problem, system boundaries and evidence needed.

  2. 02

    Build

    Use modern AI tools aggressively for implementation, research and debugging.

  3. 03

    Verify

    Test with Git, CI, browser/runtime evidence and real infrastructure.

  4. 04

    Operate

    Treat deployment, rollback, recovery and observability as part of the product.

Work with me

Need someone who can own more than one layer?

I'm currently taking on focused freelance work in web products, AI-enabled internal tools, Linux deployment and CI/CD. I'm also open to permanent engineering roles in Germany where this cross-layer experience is useful.