Skip to content

ሰላም, hello

I’m Yonas

Senior Full Stack Engineer

I fix the things that used to work, and make sure they stay fixed 🔧

Let’s talk about a role

6 years in production · Addis Ababa, UTC+3 · open to remote roles

Career Experience

Where I’ve worked and what held up because of it.

TPK Synergy Limited

Senior Full Stack Engineer

RemoteDec 2025 – Present

Amazon seller analytics across marketplace, FBA, finance, PPC and inventory. Kept 17 cron processes alive through every deploy, wired SP-API and Ads OAuth end to end, and took the platform into UK and EU marketplaces.

  • Node.js
  • PostgreSQL
  • BullMQ
  • Redis
  • Amazon SP-API
Mereb Technologies

Lead Full Stack Developer

Addis Ababa, EthiopiaApr 2022 – Oct 2025

Owned a real-money gaming platform end to end. Cut infrastructure cost ~35% (~$100K a year), took reports from 2 hours to under 30 minutes, led a software-integrity suite for a major client and mentored three junior engineers.

Full StackSeniorLead

  • PostgreSQL
  • gRPC
  • PHP
  • React
  • AWS
  • Datadog
Affiliate.com

Full Stack Developer · Contract

RemoteFeb 2023 – Aug 2023

Migrated a legacy PHP API to Node.js and tRPC while both platforms ran in parallel. Rebuilt validation in Zod, cut report time ~40%, and reached ~85% test coverage on migrated code.

  • Node.js
  • tRPC
  • Prisma
  • Zod
  • React
Eaglelion Systems Technology

Frontend Developer

Addis Ababa, EthiopiaFeb 2021 – Apr 2022

Integrated Mastercard, Visa, Telebirr and bank APIs into products with 100K+ downloads and 3B+ ETB in volume. Replaced SOAP with REST and took GTmetrix from 45 to 97.

  • React
  • Next.js
  • Tailwind CSS
  • Payment APIs

Selected Work

Real production problems, and what changed after.

17processes.
one lifecycle fix.
0102030405060708091011121314151617
CASE STUDY / ILLUSTRATED OUTCOME

TPK Synergy Limited

Keeping 17 cron jobs alive after deploy

After every deploy, recurring Amazon sync jobs died quietly. Dashboards looked fine until the data was a day old.

Outcome: Sync jobs survive deploys instead of going silent until a seller notices bad numbers.

Private production system · walkthrough on a call

BEFORE120 min
AFTER<30 min

Same million records. Less waiting.

CASE STUDY / ILLUSTRATED OUTCOME

Mereb Technologies

Reports that used to take two hours

Reporting on about a million records ran for two hours and blocked people who needed the numbers.

Outcome: Same reports, under 30 minutes. Platform performance up about 30%; infra cost down about 35% (~$100K/year).

Private production system · walkthrough on a call

ONE PRODUCT. MULTIPLE RAILS.
MastercardVisaTelebirr
CALLBACK → VERIFY → RECONCILE
SIMPLIFIED CLIENT WORKFLOW

Eaglelion Systems Technology

Nedaj payments

A fuel-purchase app had to talk to Mastercard, Visa, Telebirr, and commercial banks, with incomplete provider docs and a SOAP past.

Outcome: Payment rails that hold up in production. Live on Google Play.

Google Play
10×+
FASTER EMAIL OPERATIONSincoming_email
└─ index lookup
   └─ result
CASE STUDY / ILLUSTRATED OUTCOME

Autobridge Systems · contract

Email records that took minutes

Incoming email records on a constituent platform took minutes to load. Staff sat on the screen.

Outcome: Those operations improved by more than 10x.

Private production system · walkthrough on a call

FILE COMPARISON
SOURCE-FILE HASHING

REQUIREMENTS → TEAM + CLIENT → DELIVERY

CASE STUDY / ILLUSTRATED OUTCOME

Mereb Technologies · technical leadership

Making software changes inspectable

A major client needed a software-integrity suite with file comparison and source-file hashing.

Outcome: Delivered the suite across the full project lifecycle. Mereb’s CTO specifically credits my leadership and technical expertise in his recommendation.

Read the CTO’s recommendation
TELEGRAMCOACHING BOT

REGISTER → DAILY WORKOUT + MEAL PLAN

COACH UPDATES1:1 CONSULTATION
CASE STUDY / ILLUSTRATED OUTCOME

Villiano · contract development

Bringing fitness coaching into Telegram

Fitness customers needed daily workout and meal plans, coach updates, and access to one-to-one advice.

Outcome: A coaching workflow delivered inside Telegram, with a demonstration linked from the public project entry.

View the project and demo

Side project & contract builds

CONTRACT

Upplai

AI-powered résumé platform.

Integrated OpenAI and Gemini into résumé generation, cover letters, and ATS feedback for 500+ user profiles.

  • 500+ user profiles
View the project

Break it. Then fix it.

Pick a system, run it without the fix, then with it. Illustrative replays of real cases.

SYSTEM EXPLORERFIG. 01
INPUT → PROCESS → OUTCOME
SCHEDULER
REDIS QUEUE
WORKERS
TRY IT: BREAK IT, THEN FIX IT ▸
17cron processes

A deploy shouldn’t stop the work.

Explicit shutdown handling keeps recurring Amazon sync jobs from being stranded after a deploy.

Open the case file
INTERACTIVE SCHEMATICILLUSTRATIVE REPLAY / NOT LIVE DATA

How I Work

Same three steps, every incident.

  1. Step 1: Reproduce it

    Make the failure happen on purpose. A bug I can trigger is a bug I can fix.

  2. Step 2: Find the why

    Logs, job state, query plans. Follow the evidence until the cause explains every symptom.

  3. Step 3: Make it hold

    Ship the fix with whatever stops it coming back, then write down what happened.

Tools & Capabilities

What I reach for to build, debug, and keep systems running.

Languages
  • TypeScript
  • JavaScript
  • Go
  • Python
  • PHP
  • SQL
Backend
  • Node.js
  • Express
  • FastAPI
  • Prisma
  • gRPC
  • tRPC
  • BullMQ
Data
  • PostgreSQL
  • MySQL
  • MongoDB
  • Redis
Frontend
  • React
  • Next.js
  • Tailwind CSS
  • Material UI
Cloud & DevOps
  • AWS
  • Docker
  • Kubernetes
  • Nginx
  • GitHub Actions
  • Datadog
Integrations
  • Amazon SP-API
  • Mastercard
  • Visa
  • Telebirr
  • OpenAI
  • Gemini

Learning Journey

The foundations under the production work.

Addis Ababa University

BSc in Software Engineering

Capstone: distributed system for real-time data processing

Amazon Web Services

AWS Certified Cloud Practitioner

Certification

Ways to Work Together

Different needs, one engineer who stays until it holds.

Full-time role

Senior full-stack on a product team. From UTC+3 I overlap the whole European day and the US East morning.

Remote · Full-time

Production problem

A job that went quiet, a query that takes minutes, numbers that don’t reconcile. Bring the symptom.

Contract

System walkthrough

My production code lives in private repos. I’ll walk your team through the architecture on a call.

30-minute call

Kind Words

From a CTO, teammates, and a client. Every quote links to its source.

Recommendation letter · 31 Oct 2025
“His leadership and technical expertise were crucial in managing the project’s full lifecycle.”
Samuel Mussie · CTO, Mereb Technologies
Read the letter
“Yonas is proactive and consistently delivers high-quality results.”
Leul AyalewEngineering teammate · October 2023
Read on LinkedIn
“He goes above and beyond for solving a problem”
Mekedem GetanehEngineering teammate · August 2023
Read on LinkedIn
“His attention to detail, responsiveness to feedback, and proactive communication ensured a smooth and efficient development process.”
Analytics frontend clientCompleted Upwork contract · February 2024 · 5.0/5
Read the client review
  • Boring deploys
  • Honest numbers
  • Idempotent migrations
  • Reading the logs first
  • Fixes that hold
  • Mentoring juniors
  • Every payment state handled