Henrik Feldt

We are the operating system making second hand work. We help retailers and stores effortlessly digitise their inventory - reach out for early access at mnm.st

Currently building scientific actionable insights for marketeers via highly granular media mix modelling and clickstream data. Set up technology, recruited personell, secured investment. The majority of the time, I've been working on the product. which in October 2022 managed to through a controlled study of lowering cost of sale: https://causiq.com/blog/a-case-study-of-lowering-your-cost-of-sale-with-causiq On the tech side, we work with a lot of Python / machine learning, TypeScript, Flink, PostgreSQL/BigQuery, Hasura and NextJS. Infra on GCP/Kubernetes with a Zero-Trust network.

Assisting companies with technological change and organisational development to improve rate of and quality of delivery.

I worked together with Repath management, tech team and science team to optimise and streamline development processes, as well as put a scalable tech stack for their SaaS product in place. On the tech side, we went from manual deploys and manual database migrations to fully automated deploys and structured database migrations, as well as having a modern Python tech stack in place. I created the secure architecture for the frontend app (auth(n|z), rbac, gql, TS) and made sure it worked well with the timeseries+GIS+science backend. This including documenting and teaching, as well as hiring engineers. On the process side, we investigated different delivery models and their respective pros- and cons, iterating our way towards a model capable of making science/research work explicit, tracked and possible to interface with the higher-cadence delivery process of a modern SaaS app.

Went through the startup generator/incubator that is Antler with Causiq, pitched and received investment. Really positive experience, can warmly recommend.

When I started my own software consulting firm in 2006 I also started doing my own accouting, due to a surplus of optimism about how easy it would be. Fast-forward a little bit and I have a pile of receipts firmly nested on my desk, causing me constant worry on sight. Fast-forward a decade, and you'll find that what we have created with qvitoo is an amazing platform for making expense management and bookkeeping of receipts and invoices. By using the experience from the years I spent doing consulting, implementing Lean and doing software architecture, we have created the best system for entrepreneurs, sales and marketing to use. Complexity and clutter is moved behind smart A.I. (artificial intelligence): We make taming the complexity of the world's accounting rules our core business so you can focus on yours.

I worked with AFA Försäkring to create their next-gen platform for service delivery, both internal and external. We also built a completely new site, https://fou.afaforsakring.se as part of the delivery with a modern frontend tech stack (JAM, NextJS, React, GraphQL). As part of the position I held regular internal presentations to help spread knowledge about software architecture, system architecture and how DevSecOps can be applied successfully throughout the organisation.

I came into VOI as a ~5th engineer in late 2018, to help build the platform powering tens of thousands of scooters and hundreds of thousands of riders. At the time, the engineering team was under a lot of pressure to deliver against the hockey-stick growth numbers they were seeing and needed senior developers who could do heavy lifting. The assignment focused on technology and improving the ability for developers to deliver software more effectively. During my six months at VOI, I set up an internal dev wiki with onboarding scripts (https://github.com/haf/macos), set up a much-needed staging environment (Kubernetes, Postgres) by extracting the existing production configuration into scripts. I set up build pipelines with GCP Cloudbuild (CI) and a new deployment of Spinnaker (CD) for continuous delivery first to staging and then to prod, OAuth2-integrated to Google GSuite. I was recruiting manager for frontend developers and held bi-weekly code katas with the hired developers to improve their frontend/JS/React skills. We also set up a feature toggles-backend with Unleash which can now be used both by backend services, apps and internal dashboards. When migrating from team-by-skill to squads, I stated working with the rider-squad, where I set up a high-performance global/scalable analytics/event pipeline with logary.tech as HTTP ingestion and Flink for time-based event coalescing, Google Cloud Storage for backfill and Elasticsearch/Kibana for analytics. With detailed session-based analytics enabled, VOI can now drill down and visualise the areas needing improvement, as well as support their riders better by enabling per-user logs. There was a lot of Go-lang coding, as this is the principal language at VOI. The last week of the assignment, we completed the first successful end-to-end test of the staging environment, inclusive of the scooter, IoT-device on the scooter/GSM, our backend services, payments and the rider app!

As I worked with the micro-services architecture at Intelliplan, I needed a small web server that didn't require me to micro-manage the http-acls of Windows. Suave fit that description like hand-in-glove, but wasn't stable or fully functional, but I thought it had great potential. Website: https://suave.io Suave is now probably the most-used web library for F#. It's used in small to large organisations. I was afaik the first non-Microsoft web framework that Microsoft ported to CoreCLR/.Net Core. It's also being used for projects like FsReveal (write your presentations quick and smooth in Markdown) and as an endpoint for F# Code-Completion in Atom, the code editor from github (as a front-end to the F# compiler). Suave's guiding light is to make it easy and fun to work with web development, by being explicit and composable and thereby guiding you towards 'doing the right thing' in a functional manner. v1 was released early 2016 and v2 in 2017. In 2017 the API was also appropriated (allowed by license) by a fork which now is at the foundation for Amazon's F# SDK (in May 2018).

Tradera contracted me to be part of the team that took tradera.com from on-premise hosting to cloud hosting. The product is built around a large number of micro-services running on .Net, operated by a friendly and experienced ops team. I joined this team in January and we immediately went to work, planning and layout out the milestones we had to hit to perform the migration without more than a night worth's of downtime. Some work items include rewriting the image storage and handling to use Google Cloud Storage, another item was to create log shipping from all .Net services to Google Stackdriver via Logary Rutta. I was also part of the design discussions, provisioning and bootstrapping of VMs and k8s services. The project was successful and finished in the middle of May 2018.