Full-stack developer at Wireone Labs
A mobile screen, an API, a database shared by too many tenants, a pipeline, a Raspberry Pi on a farm, a touchscreen driver that won't behave. For three years I've followed problems wherever they go. This is how I got here.
01Origin
There's no story about a broken computer and a eureka. In 12th grade I wrote some code and liked it more than I expected to. At university the cloud sessions were the ones I didn't want to end.
Every year it got a bit more serious, and I didn't really notice it turning into the thing I'd do for a living.
12th grade
Wrote code. Enjoyed it.
University
Cloud sessions. Wanted more.
2023
First feature in production.
02First ship
My first feature in a production app was small. Notice when the phone drops its connection, tell the person, and recover quietly when it comes back. The same banner you've seen on YouTube a hundred times.
Nobody notices that banner when it works. But real people on real, unreliable networks were going to hit it every day, and it was the first time something I wrote was out there doing a job without me.
That's when it actually started. Not the code itself. Solving a problem for someone I'd never meet.
03Jul 2023 – now
I joined as a frontend developer. Nobody told me to stay in my lane, so I didn't. Since then I've worked on the app, the backend behind it, the servers it runs on, and the devices it talks to.
The chances I was given there changed how I work more than any course could have.
Every layer of the stack
Rebuilt an agritech mobile app from an empty repo. Version two draws its screens from a JSON schema instead of hard-coding them.
Moved a multi-tenant ERP from one shared database to one database per tenant. Same product, much smaller blast radius.
Replaced a separate Django-admin login on every service with one identity layer on Authentik.
Mobile builds that start from a Teams message. Grafana alerts that reach the right person before users notice.
Raspberry Pis that set themselves up through Jenkins and MeshCentral. Nobody configures them by hand any more.
Debugged Android display and touch problems at the kernel and driver level, flashing real devices to test every fix.
40%
faster responses after load testing with Locust and k6 and splitting the hot paths out
50%
less time spent on deployments once Jenkins and Docker pipelines took over
100%
tenant isolation after the per-tenant database migration
01
Documentation used to feel like homework. Now it's the first thing I do. Whoever reads it next might be a teammate at midnight, or me in six months.
02
Every problem that looked impossible turned out to be a pile of small ones. Find the first piece and start there.
03
Production goes down. Panic doesn't fix it: logs, a rollback plan and a clear head do. Then you write down what happened so it doesn't happen twice.
04The work I'm proudest of
Two online stores list the same product with different titles, photos and specs. Is it the same item, a variant, or something else entirely? I was on the team building the engine that decides. My part was accuracy, which mostly meant not trusting it.
Accuracy
Accuracy went from 73% to 81%.
01
So that “better” meant something, instead of being a feeling.
02
Every category fails in its own way. One global threshold was never going to fit.
03
A lot of “wrong” matches were really bad input. Cleaning that up moved the number more than clever rules.
04
Reviewers find mistakes faster, and every mistake they find feeds the next round of tuning.
01 / 06
Automated end-to-end provisioning, configuration and remote management of RPI’s using parameterized Jenkins pipelines.
Implemented SSO-based centralized user management replacing Django admin across multiple services and environments.
Secure storage and access control for documents with OTP login and audit trail.
Trigger mobile builds via Microsoft Teams using AWS and receive instant build links.
Real-time dashboards and alerts using Grafana and Prometheus for distributed systems.
Streamlined meeting notes with approval workflows, role-based control, and PDF exports.
05Timeline
Jul 2023 - Present
Began my professional journey at Wireone Labs where I contributed to developing robust and scalable applications, gaining hands-on experience with production-level codebases and agile methodologies.
Jan 2025 - Jun 2025
Managed end-to-end software development projects for diverse clients, from initial concept and design to deployment and maintenance.
Jul 2020 - Jun 2023
Laying the groundwork with core computer science principles and algorithms
06Toolbox
These are what I reach for today. I pick the tool for the problem, not the other way round.
Reach for daily
Used in production
Learning now
3.9
Years building for real users
22
Projects delivered
78
Features shipped
58
Technologies used in production
1718+
Contributions on GitHub
21
Repositories
07On AI
Writing code was never really the hard part. Knowing what to build, proving it's right and owning it when it breaks at 2 a.m. was the job. AI made the typing almost free. What's left is the whole job.
01
Wireone taught me to document everything. I thought that was for the next person. It turns out a model needs exactly what a new teammate needs: clear context, constraints and the why. Good docs became good prompts.
02
On the matching engine, a result only counted once it agreed with human QA. I hold AI output to the same bar: benchmark it, read the diff, try the edge case. Confidence isn't correctness, from a model or from me.
03
I use AI every day, heavily. But the decisions, the review and the 2 a.m. call stay with me. That part doesn't get delegated.
I'm not someone who knows it all. I'm someone who knows how to figure it out, and AI makes the figuring out faster.