Skip to content
View Jwrede's full-sized avatar
2 followers · 0 following

Achievements

Achievement: Pull Sharkx2Achievement: Pair ExtraordinaireAchievement: QuickdrawAchievement: Starstruck

Achievements

Achievement: Pull Sharkx2Achievement: Pair ExtraordinaireAchievement: QuickdrawAchievement: Starstruck

Block or report Jwrede

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Jwrede/README.md

Jonathan Wrede

Data & AI Solution Engineer at Possehl Digital Services, based in Münster, Germany. I build data platforms and machine learning systems and put them into production.

Website and blog: jonathanwrede.de

Work

  • Since 2027: Data & AI Solution Engineer, Possehl Digital Services. Data and AI projects for companies of the Possehl Group.
  • 2025 to 2026: Machine Learning Engineer, vilisto. Sole developer of a battery analytics platform for more than 100,000 IoT thermostats, integrated into the company Vue dashboard. Analytics warehouse with dbt and Dagster, fleet monitoring with Grafana and Prometheus on Kubernetes.
  • 2023 to 2024: Data Platform Consultant, synvert (formerly Saracus). Cloud data warehouses on Snowflake, BigQuery and Azure Synapse for five clients.

Open source

15+ merged pull requests across llm-d, OpenTelemetry and Feast, mostly CI, tests, observability and bug fixes (list).

Projects

  • tokentoll: shows in code review how changes to LLM API calls change the cost.
  • battery-passport-kit: checks data for the EU battery passport against the BatteryPass-Ready data model.
  • snni-memory-reduction: measurement harness and data of my master's thesis on the memory footprint of cryptographically secured transformer computation.
  • Anki-KaTeX-Markdown: Anki note types with Markdown and KaTeX.
  • aipreflight and llmprobe: a CI gate for AI applications and a client-side probe for LLM API endpoints.
  • llm-bench: hourly latency measurements of LLM APIs, data in llm-bench-data.

Small command-line tools, each with a short article on the blog: shutdown-blame, docker-cpg, compose-inline, envfile-compat, yq-keep, journalctl-not, galaxy-uninstall, lfs-size-track, gitignore-excludes, gh-cache-download, ubuntu-latest-check, pr-rescue, git-leftovers, diskdelta, claude-relocate, claude-rc-keep, claude-code-gc.

Pinned Loading

  1. aipreflight aipreflight Public

    CI/CD readiness gate for AI apps and LLM endpoints: evals, RAG behavior, cost budgets, observability, and rollout checks.

    Python 1

  2. tokentoll tokentoll Public

    Catch LLM cost changes in code review. Infracost for LLM spend.

    Python 4

  3. llmprobe llmprobe Public

    Synthetic monitoring and CI smoke tests for LLM inference endpoints.

    Go 8

  4. llm-bench-data llm-bench-data Public

    Open LLM API performance dataset. Hourly TTFT, latency, and throughput measurements across major providers.

    2

  5. llm-bench llm-bench Public

    Continuous open LLM API benchmark. Live dashboard, public dataset, automatic model discovery.

    Go