Technical lead and full-stack developer. I design the systems behind learning platforms used by 250,000 people a day, and help the teams who run them do their best work.
How the platform fits together: one gateway, many services, traced end to end.
I lead architecture and the hard technical calls across distributed learning platforms at Infinitas Learning. I moved live services for 250K+ daily users from Azure to Kubernetes with zero downtime, built the observability that tells us when something drifts, and work across five teams to keep quality high and the pace sustainable.
I care about clean architecture, about teams where people feel safe to say “I don’t know”, and about software that stays boring in production. Most weeks that means a mix of system design, pairing, careful pull request reviews and shipping code myself.
Before leading, I spent two years deep in the backend: the GraphQL gateway in front of 25+ microservices, OpenTelemetry tracing end to end, and frontends built on our design system.
Budapest, Hungary
daily users on the platforms I work on
250K+
microservices behind one GraphQL gateway
25+
minutes of downtime moving to Kubernetes
0
02 · Stack
Skills & technologies
The tools I reach for, grouped by where they sit in the system.
Frontend
React
Next.js
Tailwind
Material UI
Backend
Node.js
Express
GraphQL
TypeScript
Data
PostgreSQL
MongoDB
Redis
Platform
Kubernetes
Docker
GitHub Actions
Observability
OpenTelemetry
Kibana
Elastic
Testing
Jest
Vitest
Cypress
Locust
03 · Impact
Achievements & impact
Outcomes I can point to, measured where we measured them.
5+
developers led
Team leadership
Leading cross-functional development teams with agile practices and a collaborative culture that measurably raised throughput.
150%
faster deployments
Delivery pipelines
Introduced modern development practices and CI/CD pipelines, raising delivery speed by 150% across several product teams.
80%
faster releases
Release automation
Automated release note generation. Stakeholders see what shipped without asking, and releases no longer wait on manual write-ups.
45%
faster recovery
Observability
Built frontend and backend observability modules for our Node.js services with OpenTelemetry, Kibana and Elastic. Time to recover from production incidents dropped by 45%.
+65%
feature engagement
Experimentation
Set up and analysed A/B tests with Mixpanel and Snowflake. The results steered product decisions and lifted engagement on key features by 65%.
+25%
score improvement
Platform NPS
Owned the targeting rules and technical setup for measuring NPS, then used what it told us to improve the experience.
Shorter loops
from code to feedback
Internal devtools
Internal tools that sped up development and testing, tightened feedback loops and caught issues before they reached production.
04 · Activity
Contributions
A year of commits, reviews and pull requests on GitHub.
Architecture planning and technical decisions across distributed learning platforms. Mentoring, collaboration across 5+ teams, and code quality through hands-on work and thorough reviews, with a focus on team growth and psychological safety.
Core backend services for a learning platform used by students in four countries. Led the zero-downtime move of live services (250K+ daily users) from Azure Web Services to Kubernetes, added OpenTelemetry and Kibana observability, worked on the GraphQL gateway over 25+ microservices, and built frontends on the company design system.
Full-stack solutions for enterprise clients, with a focus on clean architecture, scalability and performance. Contributing to shared engineering practices across projects.
Enterprise
Node.js
React
06 · Contact
Let’s work together
Hiring for a lead or senior full-stack role? Tell me about the team and the problem. I reply to every message within a few days.
Message sent
Thanks for reaching out. You’ll hear back from me at the email you gave.
t0ms
00200 milliseconds
A short film in eight scenes
200milliseconds
The life of a single click, and the systems that make it feel like nothing.
by Botond Nuszpl · Technical Lead & Full-Stack Developer
Scroll to begin
01The tap
08:00:00.000
A student opens a laptop and taps Next lesson.
The tap becomes a request: a few hundred bytes of text asking for one thing.
GET /lessons/next
It needs to feel instant. Past about a tenth of a second, people start to notice they’re waiting.
02Distance
Light in optical fibre covers about 200 km every millisecond. Fast, but not free.
A server an ocean away would cost 60 ms there and back before any work began. So the request goes to the nearest edge, about 40 km away.
Physics sets the floor. Every design decision after this is about not wasting what’s left.
03The handshake
Before any lesson data moves, laptop and server must agree on a secret key, out loud, on a network anyone can listen to.
Think of paint. Each side mixes a private colour into a shared public one and sends the mix. Mixing is easy; un-mixing is practically impossible.
Each adds its own private colour to what it received, and both end up holding the same colour, one nobody else can make. That is the idea behind TLS.
04The gateway
The request reaches the gateway: one front door for the whole platform.
One lesson page needs the student, their progress, the lesson, its media and their grades. The gateway asks 25 services at once.
In parallel, the answer is only as fast as the slowest reply. Engineers call this tail latency. It’s why we watch the 99th percentile, not the average.
I worked on this gateway: GraphQL in front of 25+ microservices.
05Memory
Most questions have been asked before. A cache keeps recent answers in memory and replies in under a millisecond.
Only what’s new goes to the database: slower and careful, the source of truth that never forgets.
06The return
The gateway stitches the answers into one response, and it travels the same 40 km home.
184 ms
The lesson is on screen.
A blink takes about 300. The student never noticed any of it. That’s the point.
07Eight o’clock
Now do it 250,000 times. School starts, and everyone taps at once.
The services run on Kubernetes. As traffic climbs it starts more copies of each one, and replaces any that fail before anyone is paged.
I led the move of these live services from Azure to Kubernetes, with zero downtime.
08The trace
When a click is slow, how do you find where the time went?
Every hop stamps the request as it arrives and leaves. Stitched together, the stamps form a trace.
Read it like a score: the long bar is the problem. Here, the grades service.
I built our tracing with OpenTelemetry, Kibana and Elastic. We now recover from incidents 45% faster.
Nobody notices any of this.
Someone makes sure of it.
I’m Botond Nuszpl, a technical lead and full-stack developer in Budapest. I design systems like this one, and lead the teams that keep them fast, safe and boring.
Same format as before. The roles overlap because they ran in parallel. Open a span for the detail.
The root span. Everything below runs inside it.
Full-stack solutions for enterprise clients, with a focus on clean architecture, scalability and performance. Contributing to shared engineering practices across projects.
Core backend services for a learning platform used by students in four countries. Led the zero-downtime move of live services (250K+ daily users) from Azure Web Services to Kubernetes, added OpenTelemetry and Kibana observability, worked on the GraphQL gateway over 25+ microservices, and built frontends on the company design system.
Architecture planning and technical decisions across distributed learning platforms. Mentoring, collaboration across 5+ teams, and code quality through hands-on work and thorough reviews, with a focus on team growth and psychological safety.
Leading cross-functional development teams with agile practices and a collaborative culture that measurably raised throughput.
150%
faster deployments
Introduced modern development practices and CI/CD pipelines, raising delivery speed by 150% across several product teams.
80%
faster releases
Automated release note generation. Stakeholders see what shipped without asking, and releases no longer wait on manual write-ups.
45%
faster recovery
Built frontend and backend observability modules for our Node.js services with OpenTelemetry, Kibana and Elastic. Time to recover from production incidents dropped by 45%.
+65%
feature engagement
Set up and analysed A/B tests with Mixpanel and Snowflake. The results steered product decisions and lifted engagement on key features by 65%.
+25%
score improvement
Owned the targeting rules and technical setup for measuring NPS, then used what it told us to improve the experience.
The toolkit
Frontend
React · Next.js · Tailwind · Material UI
Backend
Node.js · Express · GraphQL · TypeScript
Data
PostgreSQL · MongoDB · Redis
Platform
Kubernetes · Docker · GitHub Actions
Observability
OpenTelemetry · Kibana · Elastic
Testing
Jest · Vitest · Cypress · Locust
This year
3,249 contributions on GitHub in the last twelve months.