I'm a software engineer based in Berlin, Germany.
- Building tools and libraries, mostly in Python
- Interested in AI, data, visualization, and developer tooling
- Open source contributor and maintainer
All mini testimonials made with a repository-testimonial skill.
| Project | Testimonial |
|---|---|
| ai-image-recognition-terraform | A hands-on AWS serverless walkthrough that connects Rekognition, Bedrock, Lambda, API Gateway, S3, and Terraform into a concrete image-analysis app learners can inspect end to end. — GPT-5.5 |
| asciinema-player | Extends the excellent text-native asciinema player with inline image support, preserving lightweight terminal recordings while making richer visual demos possible in documentation and talks. — GPT-5.5 |
| 👉 dirplot | Turns filesystem sprawl into something you can understand at a glance, with practical treemaps for local folders, archives, Git history, diffs, SSH/S3/Docker/Kubernetes sources, and inline terminal rendering. — GPT-5.5 |
| 👉 grokko | Turns Grok account exports into a repeatable CLI workflow, covering browser session setup, export polling, download handling, and inspection instead of leaving the process as a manual web chore. — GPT-5.5 |
| ipyrest | Brings REST API exploration into Jupyter with an interactive widget model, making requests, responses, and rich renderers feel at home inside notebooks. — GPT-5.5 |
| 👉 kata-aws-imgproc-pipeline | A serious AWS kata with staged learner stubs and a reference solution, teaching a semantic photo pipeline across S3, Lambda, Rekognition, Bedrock embeddings, DynamoDB, OpenSearch, and API Gateway. — GPT-5.5 |
| 👉 nats-primer | A clean, hands-on path into NATS: small self-contained examples spin up throwaway servers, then build from core pub/sub through JetStream, security, clustering, and microservices without setup friction. — GPT-5.5 |
| 👉 nfind | Makes file search feel like asking a precise question instead of composing a brittle command, while keeping generated Python/Node filters auditable and safely sandboxed against your real tree. — GPT-5.5 |
| obsidian-asciinema-player | Makes terminal recordings feel native in Obsidian notes, with simple Markdown code blocks for local or remote asciicasts and useful playback options right where technical context lives. — GPT-5.5 |
| obsidian-sparklines | A focused Obsidian plugin that turns tiny inline datasets or frontmatter values into theme-aware SVG sparklines, making lightweight trends fit naturally inside notes and tables. — GPT-5.5 |
| 👉 pyspark-primer | Makes PySpark approachable by turning DataFrame and SQL concepts into runnable local scripts, with the JVM/Python setup spelled out clearly enough to start experimenting instead of wrestling the environment. — GPT-5.5 |
| 👉 ray-primer | A practical on-ramp to distributed Python: focused local Ray examples progress from tasks and actors to fault tolerance, data, serving, training, and tuning without requiring a cloud cluster. — GPT-5.5 |
| 👉 skills | A useful collection of reusable agent procedures with clear install paths and validation, treating prompts as maintainable project assets instead of scattered one-off instructions. — GPT-5.5 |
| 👉 sparklines | Delivers Tufte-style microcharts where they are most useful: compact Unicode trends directly in terminals, scripts, and Python code, with support for real-world positive and negative data. — GPT-5.5 |
| streamlit-helpers | A small Streamlit prototype that sketches a practical UI surface for coding helpers, useful as an experiment in making developer utilities more interactive. — GPT-5.5 |
| 👉 svglib | A mature pure-Python bridge from SVG into ReportLab drawings and PDFs, backed by broad format tests and a CLI that makes document-generation workflows much easier to automate. — GPT-5.5 |
| 👉 time-machine-mcp-server | Exposes macOS Time Machine through MCP so backup status, snapshots, exclusions, quotas, and maintenance commands become accessible to LLM clients through a focused local tool. — GPT-5.5 |
- GitHub: deeplook
- LinkedIn: dinugherman
- X: @dinugherman