Bara Al-Sedih · MLOps Engineer · Community Builder

I run AI in production
and explain it in plain words.

Bara Al-Sedih

Bara Al-Sedih
MLOps Engineer, Palestine

See my story Message me on LinkedIn

01 No map

When I started, there was no map.

The field was new in Palestine. No communities, no company trainings. I learned it the hard way: a lot of mistakes, and a lot of hours spent on the wrong choices.

02 Practice

So I practiced wherever I could.

Programs, internships, hackathons, and freelance work. One stop led to the next.

  1. May · Oct 2024

    Machine Learning Program

    Udacity

    Classification, regression, and clustering. Neural networks and hyperparameter tuning.

  2. Jun · Aug 2024

    Backend Developer

    Freelance team

    RESTful APIs with NestJS and TypeScript.

    “You picked up NestJS in no time.” Yazan Faroun
  3. Aug 2024

    1st

    DataBridge Hackathon

    Gaza Sky Geeks and DataCamp, Ramallah

    Our team, Legends, finished top of the board.

    DataBridge Hackathon leaderboard with team Legends in first place
  4. Oct 2024 · Oct 2025

    Student, DataCamp scholarship

    Gaza Sky Geeks

    I joined the scholarship as a student, and spent a year learning data on DataCamp.

  5. Jan · May 2025

    Three internships

    RYDLR Cloud Services · Udacity · PalSoft

    Machine learning, data science, and deep learning: models in TensorFlow, Scikit-learn, and PyTorch, SQL on real data problems, and CNNs and YOLO for computer vision.

  6. Feb 2025

    1st

    Stichting SPARK 3rd Hackathon

    Udacity graduates

    We built a mental health chatbot on a fine-tuned Llama model, and added Arabic conversations to the dataset.

    The screen announcing the Palestine Launchpad hackathon win
  7. Jun 2025 · now

    AI Engineer

    Freelance

    I delivered an AI Agents training for a client. It turned into an ongoing product build, and two production AI agents.

03 The real thing

Then I landed my first job in the field.

That’s when I promised myself I would help grow this field in Palestine.

Sep 2025 · now · MLOps Engineer · one story from the job

GitHub Actions couldn’t reach our staging server. So I made the server reach GitHub instead.

GitHubrelease tags
Private servera worker runs here

Push a tag. The server deploys itself.

No inbound access needed. No manual deploys.

  1. A worker on the server checks GitHub for new release tags.

  2. It spots when a new version is out.

  3. It deploys locally, on the server.

Now I run AI in production. I own a large part of my team’s infrastructure, deployments, and production operations.

  • I turned one server into the place where the ML, LLM, and data analysis teams build, test, and deploy.
  • Releases after the first setup: config, validation, deploy, health checks, and rollback.
  • A model-serving service that several products use. I built it and I maintain it.

Outside work · Atlas AI

I trained a model. Then I built an entire MLOps platform around it.

Atlas AI watches ML models and LLMs after they ship. It is open source, built across three repos, and covers observability, RAG, and CI/CD.

  • Detects drift, when live data stops looking like training data.
  • Tracks model performance over time.
  • Traces every LLM call.
  • Raises an alert automatically.
  • RAG over Obsidian notes

    Hybrid search, cross-encoder reranking, and LLM traces sent to ClickHouse.

    Code
  • Real-time threat detection

    A data pipeline, several ML/DL models with one decision layer, and a Dockerized REST API.

04 Giving back

Now I draw the map I didn’t have.

I learned the hard way, so I make it easier for the next person. I lead, I mentor, and I write.

I lead · Gaza Sky Geeks

I was a student in the DataCamp scholarship.
Now I lead it.full circle

30 volunteers. Students from 7 universities. More than a year of it.

  • Mentor and guide learners
  • Arrange online and on-ground gatherings
  • Organize technical and soft-skills sessions

I mentor · Jun 2025

One hackathon later, I went back as a mentor.

1st+ 2nd

At the Stichting SPARK 4th Hackathon I co-mentored two teams out of 24. ThemeSolve took 1st place. Hirelyze took 2nd.

Cover slide of ThemeSolve, the team that took first place
“Bara is driven, collaborative, and deeply committed to making an impact, it’s been inspiring to watch his growth :)”

I write · LinkedIn · three topics

I won’t hesitate to share every useful lesson I learn.

AI in productionGitHub Actions couldn’t reach our staging server.

Real incidents from my own systems.

  1. Built a production-grade RAG system over my Obsidian notes.

  2. I trained a model. Then I built an entire MLOps platform around it.

  3. Owning the infrastructure is part of the MLOps job.

People and careerTwo Saturdays, two events, 250+ students in the room.

Lessons from leading volunteers, mentoring, and my own path.

  1. Two years ago, I started my journey in AI and ML.

  2. Two universities, one room, and one goal: keep our students learning.

  3. Getting rejected doesn’t mean the connection is over.

Developer craftHaving a GitHub Portfolio is one of the most essential things for developers.

Git, Python, the terminal, and the AI tools I use.

  1. Stop using Pandas for your production pipelines!

  2. Running pipelines shouldn’t feel like decoding a spell.

  3. Helpful trick I learned recently for GitHub.

05 Next

2,800+ people follow along

It’s still the beginning.

If you’re learning data or ML and feel stuck, message me. If you run ML in production, I’d like to compare notes.

For the record

The paperwork behind the story

Certificates

7 selected certificates, by who issued them

Education and languages

Computer engineering, in Nablus

Bachelor of Computer Engineering

An-Najah National University, Nablus · 2020 to 2024

Two graduation projects: a tile grout filling robot, and FreeDevelopersCamp, a learning platform.

Arabic

Native.

English

Professional. C1 Advanced on the EF SET.

Toolbox

What I use to run ML in production

Serve and ship

Model serving · Model packaging · FastAPI · Docker · CI/CD with GitHub Actions

Monitor

Model monitoring · Data drift detection · Weights & Biases

Run

AWS · Python · Git · PostgreSQL · Makefile · Poetry