GET /ward 200 OK —
personal website, instrumented

Ward Bekker_

Most of my career has been an argument about making technology genuinely useful. I started as a software developer building high-traffic websites, then spent years in solutions engineering on complex data platforms at Hortonworks, Cloudera and Grafana Labs — which taught me as much about customers as about systems. I'm now a Senior Principal Product Manager at Grafana Labs, working on product-market fit and adoption across our cloud products, particularly the newer AI offerings. I never stopped being a builder; most of that is in the trace below.

Career trace

the record

Every bar above, in full

A CV flattens a life into a list. This is the same information drawn the way a tracing system would draw it: spans that start, overlap, and sometimes keep running in the background long after the job that spawned them has ended. The overlaps are the interesting part.

00 · ward.bekker · 2000 → now · 26.9y

One long-running request

Started somewhere around 2000 and still in flight. Handles concurrency poorly, recovers well from failure, and has never once been the sole responder to a P1.

The children overlap because that is what actually happened. Some of them are still running.

status
200 OK, no timeouts
retries
several, all instructive
region
Netherlands, CET/CEST

01 · build.websites · 2000–2015 · 15.8y

Software developer, high-traffic websites

Fifteen years of building and running web software: .NET at dash-mp and Tam Tam, R&D at Radventure, then a long stretch at TTY Internet Solutions as technical lead and later product owner. Sites that got more traffic than they were designed for, back when "web scale" was still a conference talk. Intermediair was the big one.

Also got Erlang builds working on Travis CI in its early days, when CI was a novelty rather than a line item. A small patch and a long argument about versions.

The lesson that stuck: measure before you optimise, then measure again. Same lesson twenty years later, better dashboards.

stack
PHP · MySQL · .NET · Erlang
lesson
the database is the bottleneck

02 · search.erlang · 2011–2012 · 1.0y

A search engine, in Erlang

Wrote a full-text search server for a tutorial series: indexing, stemming, relevance ranking, faceting, async results and websockets. Later turned the WAND / max_score top-k algorithm into a small library.

This is where I fell for the BEAM. Processes that fail alone are a very calming thing to work with.

stack
Erlang · OTP · websockets
algorithm
WAND / max_score top-k

03 · mobiquity.se · 2015–2016 · 1.0y

From building to sitting with customers

Solutions Engineer and Solutions Architect at Mobiquity, working on digital engagements for large brands. The first role where the job was less about writing the code and more about understanding what the customer was actually trying to do.

A short span, but the hinge: after fifteen years of building, I started spending my days in customer conversations instead of in an editor.

shift
builder → customer-facing
lesson
the problem is rarely the stated one

04 · hortonworks.se · 2016–2019 · 3.1y

Solutions Engineering at Hortonworks

Three years sitting with customers who were trying to make Hadoop, Spark and Kafka do something useful. Complex data platforms, a lot of it in big-data security: pipelines ingesting very noisy data, and a contribution to Apache Metron along the way.

Learned to read logs for a living, which turned out to be the whole job later. Most of the work was convincing a system to be honest about what it had just done.

stack
Hadoop · Spark · Kafka
domain
network security
superpower
grep

05 · cloudera.se · 2019 · 0.8y

Cloudera, briefly

Hortonworks merged into Cloudera in January 2019, so the last stretch of that work happened under a different logo. Two platforms, two customer bases, one roadmap. The overlap with the Hortonworks span is not a drawing error.

Short span. Useful lesson in how much of a product is really just people agreeing on what the words mean.

context
post-merger
lesson
integration is a product problem

06 · grafana.se · 2019–2025 · 5.4y

Solutions Engineering at Grafana Labs

Joined in 2019 as a Senior Solutions Engineer and spent the next five years as the technical person in the room during customer pilots: Senior SE, Lead SE, Senior Principal SE, and finally Senior Principal SE in the Office of the CTO. The job was to answer the hard question, then carry the answer back to engineering.

Loki was the obsession: a pull request into Loki, the LokiLogger Elixir backend, workshops, and a pile of YouTube tutorials made because writing the docs once was faster than answering the same question forty times.

focus
Loki · observability
output
demos, workshops, videos
arc
SE → Senior Principal SE

07 · grafana.product · 2025 → now · 1.8y

Product, still customer-first

Now Senior Principal Product Manager for Cloud Solutions in the Office of the CTO at Grafana Labs, focused on product-market fit and adoption across the cloud products, particularly the newer AI offerings.

The job is still a customer conversation, just earlier and with more diagrams. A lot of time with customers, spotting opportunities, and working with product and engineering to shape what gets built.

Those solutions-engineering years are the advantage: I have seen what happens when a technically impressive product meets a customer who cannot see why they would use it.

scope
cloud solutions · AI offerings
focus
product-market fit, adoption

08 · gcx · 2025 → now · 1.5y

gcx: the side project that got promoted

gcx is a Grafana CLI optimised for AI agents. I started it as a side project because I saw an opportunity and wanted to explore it. Using AI coding tools, I built the first version. It gained traction, attracted a growing team, and became an important part of Grafana's AI direction. I'm still involved in shaping where it goes.

Public preview in April 2026, generally available that July. This is the span I would point at if someone asked why I still call myself a builder.

started
side project
became
team + product
status
GA, July 2026

09 · ai.tooling · 2023 → now

AI brought the excitement back

AI has brought back a lot of the excitement I felt when I first started developing software. I use LLMs and agents extensively, and I love how they make it possible to go from an idea to something working much faster than before.

It is the same loop as always — understand the problem, figure out what's possible, build the thing — just with a much shorter distance between the thought and the prototype.

stack
LLMs · agents · AI coding tools
effect
idea → working thing, faster

10 · side.quests · 2011 → now

Side quests

Things built because they were interesting: an Erlang client for a NeuroSky EEG headset, a d-gap / Elias-gamma compression library, a Top-K search implementation, and whatever the weekend allowed.

Also a few years helping run Cross Functional AMS, a meetup in Amsterdam for people using functional languages commercially. The point was not the language, it was meeting people with real production scars who would talk about them. The group is still running, now as FP AMS.

None of these shipped a roadmap. All of them made the day job easier, eventually. One of them got promoted.

pattern
interesting > planned
outcome
one got promoted

11 · life.outside · 2000 → now

Everything that isn't work

A wife, two kids, and a preference for places with unreasonable amounts of nature. Long-distance running when the knees agree — and as often as possible somewhere new, because work takes me to a lot of cities and a city at six in the morning is a different city.

The other half of that is looking at light. Coasts, cedar forests, ridgelines, weather doing something dramatic over water. Mostly badly, and mostly happily.

Not a blocker, but always running. It is why the trace has gaps on sunny weekends, and why the rest of it works.

status
non-blocking, high priority
location
between Amsterdam and The Hague
the other trace

Where it runs, and what the light did

Two things that never made it onto a roadmap. The route below is the recurring one; the times are the ones worth keeping. No photos — the mornings are better as a log.

run.world · cities at 6am

  • AMS Amsterdam The Amstel at six, before the bridges fill.
  • LON London The Thames path at low tide, city glinting behind you.
  • SFO San Francisco Bridge in fog, then the Presidio climb. Hills hurt.
  • YVR Vancouver Stanley Park seawall. Cedar, rain, herons. Never once dry.
  • SIN Singapore Marina Bay at 5:30, 29 degrees, already sweating.
  • TYO Tokyo The river path at dawn. Every runner nods. Nobody talks.

field.notes · mornings worth keeping

  1. singaporemarina bayheat haze on the water, no wind at all
  2. vancouverstanley parkcedar and rain, herons, no usable light
  3. tokyoodaibathe bay at low tide, gulls, everything grey
  4. amsterdamamstelflat light, four degrees, no photo
  5. san franciscolands endfog so thick the bridge was a rumour
  6. londonrichmond parkdeer in mist, the one that got away
trace log OK 12 spans · one file · no dependencies