Furkan Doğruer · Software Engineer

Hi, I'm Furkan.
I build software.

I build Python backends, TypeScript/React applications and practical AI tools in Munich. I use coding agents, then review and test the changes.
Explore Work
Download Resume PDF
GitHubLinkedIn

What I work on.

AI tools & automation

Code-aware retrieval in DevGraph; browser workflows and explicit controls in Job Agent.Explore AI work

Backend & applications

Python APIs, React interfaces and relational data in professional and personal work.See my experience

Code, review & testing

Coding agents support implementation. Review, debugging and tests help me check the result.How I work

Selected work

Personal projects, with their current limits.
Repository-aware RAG
Conceptual architecture · original illustrationKeep the source in the context.
Prototype · public source

DevGraph AI

Ask questions about a local TypeScript codebase and inspect the source behind the answer. A FastAPI, Next.js and Qdrant prototype.
Python / FastAPI
Next.js / TypeScript
Qdrant
Gemini
Case studyPublic source
Controlled automation
Conceptual architecture · original illustrationProgress needs a known outcome.
In development · private source

Job Agent

A personal project for job discovery and browser workflows, with saved state, bounded retries and human handoffs. Still in development.
Python
Playwright / CDP
SQLite
Cursor / Codex
Case study
Simulated telemetry
Conceptual architecture · original illustrationFrom readings to visible state.
Simulated demo · private source

TelemetryGuard AI

Simulated telemetry, rule-based alerts and a live-updating dashboard. Built with FastAPI, Next.js, SQLite and WebSockets; reports use mock templates.
FastAPI
Next.js / TypeScript
SQLite
WebSockets
Case study
TWO COMPLEMENTARY PRACTICES

AI in the product. AI in the process.

AI as part of the application

DevGraph connects source code, embeddings and retrieval to an LLM response. The sources are visible; answer quality remains unevaluated.Inspect the chunking lab

AI as part of the engineering process

Job Agent separates coding-agent implementation, review and acceptance checks. Generated code still needs inspection.Read about the workflow

From task to checked change

  1. 01 · PlanClarify requirements, constraints and acceptance checks.
  2. 02 · ImplementUse coding agents in isolated worktrees; connect the components.
  3. 03 · ReviewInspect the diff, assumptions and reviewer feedback.
  4. 04 · TestRun relevant checks and review the tests themselves.
  5. 05 · ValidateCompare behavior with requirements, debug gaps and decide what is ready.
Coding agents help with the work; I remain responsible for the result.

Have a project or role in mind?

Explore my background or get in touch about software and AI engineering.
Get in touch
View CV