Adway

- 39 ATS analysis & reporting to clients; improving renewal success by 70%
- Automated monthly sales reports with Python, saving 15 hours per month. 
- On our way to 100M ARR in 2025, leveraging data and tech.
- Tech stack: M365, Excel, Python, PowerPoint, Loom

Data Scientist & BI Consultant

Alterfy

Advising the software and data team in building a robust SaaS web app leveraging AI,  analytics, and data. Perform data analyst and data scientist tasks by measuring KPIs and building machine learning models.

AI & Tech Advisor

ROCK

- Led the BI using tools like BrightGauge, PowerBI, and HaloPSA to create and update over 200 reports and dashboards.
- Managed the data transition for an MSP used internally called HaloPSA which delivers a 5x growth to the company.
- Data engineering, and data science work on Microsoft Fabric (pipelines, Lakehouses, Warehouses, Notebooks...)
- Data consulting in data projects for clients.

Data Scientist

Kadogoodiz

Head of data science and analysis with the goal of fetching, cleaning, analyzing, and reporting data to senior partners.

Uhubs

+ Analyzing the data sets to extract insightful information (Python, Pandas).
+ Cleaning and feature engineering on data sets.
+ Integrated natural language processing (NLP) to analyze textual data (nltk, stemming, tokenizing).
+ Visualized insightful information (Matplotlib, Seaborn, Plotly)
+ Deep research on benchmarking to increase sales workflow.
+ visualizations and reporting

Consultant Data scientist

Riad coaching

Performance Coach

Setexam – The Agar Specialists

+ Maintained databases using MySQL
+ Handled part of the AWS infrastructure and migration from on-premises to the cloud
+ Cleaned data prior to training and analysis using Python (Pandas, NumPy)
+ Visualized important metrics and outputs using Matplotlib and Seaborn
+ Been part of the machine learning team and made sure to apply best practices and optimize the code.

Assistant data scientist

+ Designed maintenance software ‘Coswin 7’ and trained technical staff to use it for preventive and corrective maintenance.
+ Trained a newly hired maintenance technician to manage and use Coswin 7 for troubleshooting and effective use.
+ Automated part of the process concerning the dosing and transfer of chemicals for cleaning purposes using Allen Bradley API language.
+ Supervised and visualized in real-time the automated process via RSLogix 500.
+ Worked with IO-Link, a modern technology of sensor/actuator that was added to the company’s workflow.

Engineer Intern

Syndic8

Collective created with a bunch of friends in 2015, eager to grow the Moroccan musical scene. 
As a co-founder and musician I dealt with :

+ 10+ High-quality events all over Morocco
+ Booking 16 International renowned artists. 
+ Budgets ranging from 1000-13500£.
+ Communication over Facebook and Instagram + Ads
+ Performing live as a professional around Morocco and France

Co-Founder

+ Exploring the industrial world, API's, sensors, programs, servers, machines, supervision, maintenance...

Stagiaire en ingénierie industrielle

🤖Data Scientist | Turning Raw Data into Money Generating Insights | Data-led Growth for Your Business 🚀

Freelance Data Scientist | ML Engineer | SQL | Python | Power BI | MLOps

Data analytics and Machine learning models built 5X faster - no quality compromise. Worked with 14 clients in 7 different industries in the last 3 years. Helped current company scale to $5M+ ARR. Featured TOP RATED PLUS talent on Upwork in 2025 with 100% job completion and satisfaction.

My Toolkit:
- Data analytics: Python, SQL, Excel
- ML Engineering: FastAPI, AWS, Streamlit, great_expectations
- Data visualization: PowerPoint, PowerBI, Plotly, Matplotlib
- Data modeling: Scikit-learn, TensorFlow, Keras
- MLOPs: CI/CD (GitHub Actions), Docker
- Additional tools: Notion, n8n, Canva, Git/GitHub

My Stats:
- 5-stars reviews on 99% of projects delivered.
- 100% Job Success Score for on-time, budget delivery.
- Helped scale currently business to $5M ARR with ATS analysis & Reporting.
- 1 year TOP RATER PLUS on Upwork.
- 1200 hours billed.
- $40k+ earned.

What You’ll Get Working With Me:
- A data analyst who is a real problem solver.
- Quick responses (within the hour on working hours)
- Clean, maintainable code and detailed documentation
- A clear “NO” if something isn’t feasible; no wasted time
- A well-crafted report or dashboard as per the requirements

My clients think:
- "Anas is great and highly knowledgeable. He is passionate about solving data science challenges and has helped me significantly streamline some of my processes, enabling better scalability.
- "Anas, was prompt and thorough and I would def use him again in the future."
- "Anas is great to work with. He made the job so fluid and his knowledge of data science makes him someone worth listening to. I’m glad I chose him for my small business. I’m looking forward to what’s coming next, and I would definitely hire him again."

Popular Clients & Products:
- Uhubs: Financial company based in London
- AllAboutGroup: Law recruitment firm based in London
- Growth Dynamics: Marketing and data strategist based in London.
- Adway: Tech and recruitment company based in Gotheborg, Sweden

About me:
- Football fan since I can remember (Real Madrid fan)
- Tennis lover and practitioner (one-handed backhand style)
- Mangas fan (One Piece, Naruto, Bleach)
- Travel addict (South Asia, Europe, Africa, America (soon))
- Reading is something I enjoy as much as coffee (oat latte fan recently)

-----

YouTube Channel
→ GO HERE: https://www.youtube.com/anas_riad

Request services via LinkedIn or Email below:
r.riadanas@gmail.com

Anas Riad

The ML project you wish you built sooner.
2.5hr of hands-on learning ⤵️

No excuses for skipping end-to-end ML now! 🤣

(PS: Link to the full YT video in the comments)

When I first started working on churn prediction, 
I noticed something that always bothered me.

There are tons of Regression projects out there, 
but 𝟗𝟗% 𝐨𝐟 𝐭𝐡𝐞𝐦 stop at the notebook.

It’s like drawing the blueprint for a house but never building it.

So I created a full end-to-end regression project that goes from:
raw dataset → trained model → deployment → tests → cloud.

Here’s what I cover in this project & video:
→ Split data by date (time-series aware)
→ Data Quality with Great Expectations
→ Well-structured pipelines (feature, training & inference)
→ Unit & smoke tests to keep things reliable
→ Experiment tracking with MLflow
→ Containerization with Docker
→ FastAPI serving layer
→ CI/CD with GitHub Actions
→ AWS (ECS, ECR, S3, ALB) for Deployment

It’s the same type of stack I use in freelance + personal work, but applied to regression.

Looking forward to your thoughts on it.

🌟 Link to the full YT video in the comments

PS: Would you be interested in a dedicated hands-on MLE platform? 🤔

Data Scientist & BI Consultant @Adway • ML Engineer • YouTube @ Anas Riad

The ML project you wish you built sooner.
2.5hr of hands-on learning ⤵️

Thanks to Miguel Otero Pedrido for inviting me and supporting my content (and the sleek diagram!).

The video has been live for a few weeks.
Now, the full article is available on The Neural Maze Substack.

This gives you a chance to understand the flow better.
Then you can combine it with the video and hands-on.

Let’s build a network with real builders and add value to each other!

📌 Link to the article and video in the first comment

Most people get stuck between learning ML and deploying it.
Here’s the roadmap I wish I had ⤵️

PS: Link for the full YT video below ⭐

— — —

1️⃣ Python Programming
→ The foundation. Everything in ML starts here.
→ clean code, functions, data handling, and reproducibility.

🔗 Python Programming with Mosh (4hours): https://lnkd.in/e_y5APBb
🔗 100 Days of Code: https://lnkd.in/eHQn6HYk

— — —

2️⃣ Math & Statistics for ML
→ Build intuition for algorithms.
→ You can’t optimise what you don’t understand.

🔗 Statistics Fundamentals w/ StatsQuest (playlist): https://lnkd.in/ef3rg2UP
📕 Book: Practical Statistics for Data Scientists

— — —

3️⃣ Machine Learning Fundamentals
→ Learn how models think: data prep, feature engineering, model selection, and evaluation.

🔗 Stanford ML: https://lnkd.in/epviQVc8
🔗 ML specialization (coursera): https://lnkd.in/edbCT-3B
🔗 All ML fundamentals (short overview): https://lnkd.in/eSdDdMWS

— — —

4️⃣ Deep Learning Essentials
→ Understand how neural networks learn and train, from simple perceptrons to CNNs and Transformers.

🔗 MIT intro to deep learning (playlist): https://lnkd.in/ebS2_kvx
🔗 Understanding Deep Learning (notebooks): https://lnkd.in/eUyS-ggD

— — —

5️⃣ MLOps
→ The most in-demand skill today.
→ Automate, version, deploy, and monitor ML systems like a pro.

🔗 Docker for beginners: https://lnkd.in/eQXFqutQ
🔗 FastAPI for beginners: https://lnkd.in/eg7xnhsY
🔗 MLFlow tutorial: https://lnkd.in/enjdQ_zt
🔗 GitHub Actions tutorial: https://lnkd.in/etYr8DDf

— — —

6️⃣ End-to-End ML Projects
→ Bring it all together. Real data, real models, real deployment.
→ This is what turns knowledge into experience.

🔗 Get started with AWS: https://lnkd.in/e98PMn6d
🔗 ML Churn End-2-End (1.5 hours): https://lnkd.in/e3Fmdh3t
🔗 ML Regression End-2-End (2.5 hours): https://lnkd.in/ebMWGweb
📕 Book: Designing machine learning systems

— — —

Happy learning!

🎥 Full video: https://lnkd.in/ekMz9MF8
💬 Comment below: which stage are you currently in?

7 steps to setup Docker for data science projects

(90% of beginners get confused)

1– Choose a base image
→ Start from a lightweight Python version (e.g. python:3.11-slim).

2– Set a working directory
→ Define where your code will live inside the container.

3– Copy dependency file
→ Add only requirements.txt first (helps with Docker caching).

4– Install dependencies
→ Upgrade pip and install your project libraries.

5– Copy project files
→ Bring in your source code and scripts.

6– Expose the port
→ Make your FastAPI or Flask app accessible (commonly port 8000).

7– Define the CMD
→ Tell Docker how to run your app (e.g. start Uvicorn server).


💡 Extra Docker tips for data science
→ Build the image → docker build -t project-name .
→ Run the container → docker run -p 8000:8000 project-name
→ Docker Compose → Manage multi-container apps (API + DB), add volumes to persist data
→ Docker Build Cloud → Up to 39× faster builds for large projects
→ docker init → Auto-generate Dockerfile, compose file, and ignore rules
→ Separate your dev and prod containers


📌 Save this for your next project.
♻️ Share this to help your network in data.

9 reasons every data scientist should read 
"Designing Machine Learning Systems" by Chip Huyen

1– It pushes you to design systems, not just models
→ The book shows that ML is more than training models; 
→ it’s about building data pipelines, serving layers, and monitoring loops.

2– You get a holistic, iterative framework
→ ML is not linear. Huyen frames it as an evolving cycle:
collect → train → evaluate → serve → monitor → improve

3– Design decisions get context
→ Every technical choice (data format, retrain frequency, feature logic) is explained in terms of trade-offs and system goals.

4– It’s written for engineers, not just theorists
→ You’ll find architecture diagrams, deployment practices, and design principles, not just equations.

5– You learn how to scale beyond toy models
→ Covers how to move from small experiments to production-scale systems, including distributed training and monitoring.

6– Case studies ground the lessons
→ Real-world examples from tech companies make each concept practical, not abstract.

7– You’ll understand data pipeline engineering
→ From ingestion to streaming and transformation, Huyen explains why most ML challenges are actually data challenges.

8– Monitoring, model drift & updates are core topics
→ How to catch drift, retrain automatically, and measure if your system still meets business goals.

9– You’ll think in systems, not notebooks
→ Encourages you to view ML as an ecosystem: feature stores, experiment tracking, deployment, and serving all connected.

———

My top 3 learnings from the book:
1– Split time-based data by date, not randomly
→ I enjoyed the data leakage section and learned to split by date in time series and ALWAYS train on the past and test on the future, just like real deployment.

2– Focus on business constraints first
→ The “system mindset” chapter changed how I frame projects. Latency, throughput, and cost matter as much as accuracy. Sometimes a 95% accurate model that runs in 50 ms is better than a 99% one that runs in 5 seconds.

3– Think distribution, not just average
→ A single performance metric hides what matters. The model must handle skewed data, seasonal trends, and outliers because real-world data never looks like your training set.

———

If you’re building your first serious ML project, this book will shift how you think about data, design, and business impact forever.

PS: Have you read it?
♻️ Share this to help your network in data.

6 reasons ML engineers should use GitHub Actions

(90% miss number 5)

1– Automate your workflows
→ Run tests, lint code, and build containers automatically whenever you push to GitHub.

2– Keep models reproducible
→ Ensure each commit trains, evaluates, and logs results in the same controlled environment.

3– Simplify CI/CD
→ Build, test, and deploy models or APIs straight from GitHub, no manual steps needed.

4– Integrate with cloud & containers
→ Trigger deployments to AWS, GCP, or Azure, or push Docker images to registries without leaving your repo.

5– Catch issues early
→ Run pytest, data validation, and static checks before merging so broken pipelines never hit main.

6– Collaborate with confidence
→ Review every change with automated checks, artifacts, and logs attached to each pull request.


A few things to know about GitHub Actions:
✅ Hosted on GitHub
✅ Configured with YAML
✅ Easy to set up & maintain
✅ Public repos → Free (with limits)
✅ Private repos → Some free mins, then paid
✅ Limited by GH infra (or use self-hosted runners)

—

GitHub Actions turns your ML repo into a self-running machine.
Train, test, and deploy seamlessly with every commit.

📌 Save this for your next project.
♻️ Share this to help your network in data.

Your Data Science interview is won in the first 15 min
20 questions you must be ready to answer ⤵️ 

Foundations
1. What’s the real difference between correlation and causation, and how do you prove it?
2. Explain bias vs variance as if you’re teaching a new analyst.
3. What are common data leakage examples, and how do you spot them early?

Data Cleaning & Feature Engineering
4. How would you handle missing data differently for categorical vs numeric features?
5. When is scaling not needed, and why?
6. How do you detect multicollinearity and decide what to drop?

Modeling
7. Logistic Regression vs Random Forest: how do you pick one for a real business problem?
8. Explain the ROC curve in plain English. What does AUC actually mean?
9. How do you know your model is overfitting without looking at code?

Experimentation & Evaluation
10. Walk me through how you’d design an A/B test for a pricing change.
11. What metrics would you use for imbalanced classification?
12. How do you deal with data drift after deployment?

MLOps
13. How would you track experiments and compare model versions?
14. What does a good CI/CD setup for ML look like?
15. How do you monitor model performance in production without real labels?

Deployment
16. Walk me through deploying a model with FastAPI + Docker.
17. What’s the difference between batch and real-time inference pipelines?
18. How do you handle model rollback safely?

Communication & Impact
19. You built a model that’s 10% more accurate, how do you explain its value to a non-technical exec?
20. Give an example of when your model was right, but the business didn’t benefit, what went wrong?

⸻
Can you answer these clearly? If not, that’s the gap to fill.

Tip: Record yourself answering these and rewatch. You’ll quickly spot what needs work. Rinse and repeat.

Use the cheatsheet as a starting point, not as an absolute response.

📌 Save this for your interview prep.
♻️ Share this to help your network in data.

Everyone wants to be an AI Engineer.
But not many want to master the foundations first.

Miguel, a Senior ML Engineer, recently told me:
“AI engineering becomes easy when your MLE foundations are strong.”

Still, most people skip the basics like:
→ CI/CD pipelines
→ Feature engineering
→ Model experimentation & tracking
→ Deployment, monitoring & observability
→ Reproducibility (Docker, version control, etc.)

Then they open n8n,
build a few workflows,
and call themselves AI engineers.

I’m not here to tell anyone what to do.
But over the years, I’ve learned to tune out the hype and stick to my plan.

The AI engineers you admire on LinkedIn?

Most have 5+ years of experience in data science and ML engineering.
They didn’t shift into AI overnight.

Stop following the hype.
Start building your career the right way.

77% of data scientists use Pandas for EDA 

Here are 10 tips you need to master today

(PS: This post is focused on pre-analytics)

1️⃣ Use the right function: read_csv, read_excel, or read_sql depending on your source.
2️⃣ Skip unnecessary header rows to start from the actual column names.
3️⃣ Check your file’s encoding (utf-8, latin1, etc.) to avoid weird characters.
4️⃣ Parse your date columns with parse_dates and convert timezones if needed.
5️⃣ Clean numeric columns with commas or symbols before converting to float.
6️⃣ Define missing values properly (na_values=["", "N/A", "-"]).
7️⃣ Remove duplicates with drop_duplicates().
8️⃣ Reset your index after filtering or merging.
9️⃣ Convert mixed types using pd.to_numeric(..., errors="coerce").
🔟 Always inspect data before analysing: df.info(), df.describe(), df.head()

Most analytics still run on tabular data, spreadsheets, CSVs, SQL tables.

Getting these small steps right saves hours of debugging and avoids silent errors that ruin your results.

Do you use Pandas for data analytics?

📌 Save this for your next project.
♻️ Share this to help your network in data.

90% of data scientists who reach out to me…
Get stuck on one thing 👇

→ How do I build the “best” portfolio?

It’s a question that sparks endless debate.

Some say, “Do as many projects as possible to get practice.”
Others say: “Do fewer projects, but go deep.”

Here’s my take: both work.

If you’re a beginner:
→ do more to build momentum.

Once you have a base:
→ go deeper to show mastery and value.

But there’s one thing almost nobody talks about:
Pick projects based on your own interests.

That changes everything:
→ You’ll actually enjoy the work.
→ You’ll put in the hours without forcing it.
→ You’ll understand the business side better.
→ You’ll finally answer the questions you always had.

PS: Have you tried building a portfolio this way?

🌟 Full episode with Miguel on YT, Link in comments

2 years ago, I pivoted to data science.
I had zero traction.

Today, I reach over 1M people a year.

(Read till the end. I’ll share what actually worked 👇)

I could say it’s easy.
But it’s not.

I started writing on LinkedIn in Jan 2022.
Bought Justin Welsh’s course.
Learned copywriting.

I talked about productivity, then coaching, then data science.

Each time I pivoted, I lost followers.
But I gained friends, real ones.

People who supported me, challenged me, and helped me grow.

LinkedIn has been:
✅ a place to meet amazing people
✅ a place to find inspiration & accountability
✅ a place that forced me to learn and share more
❌ also a place that made me doubt myself at times

Over time, I learned to keep the good and ignore the bad.

So… how can you grow faster (and maybe do even better)?

1️⃣ Stick to ONE topic and go deep
Fluffy content gets likes, not authority.

2️⃣ Be consistent
Consistency makes people:
→ remember you
→ look forward to your posts
→ recognize your tone & style

3️⃣ Join a circle of creators in your niche
→ LinkedIn boosts posts that get engagement from relevant people
→ Genuinely support others, they’ll do the same
→ This alone can 10x your visibility

4️⃣ Don’t compare yourself to others
→ LinkedIn takes time
→ It’s full of noise
→ Follow people who inspire you, and mute the rest

If you can reach 1M people a year in your field, for FREE...
do you really think LinkedIn is a bad investment?

International students, do this before moving to the UK 🇬🇧
(90% of students miss tip #3)

1– Understand your visa options
→ Know the difference between Student, Graduate, Skilled Worker visas, etc.
→ On a student visa, you’re allowed 20 hours of weekly work
→ On a graduate visa, you can freelance, work, or start a business
→ Check working hours allowed, duration, and switching rules early; it’ll save you stress later.

2– Explore part-time and freelance work
→ Many students earn through part-time jobs or freelancing.
→ Start browsing options (e.g., Upwork) before you arrive.
→ I was sold the idea that I could find a part-time job in my field easily (with my engineering background), but it was far from the truth. Do your research.

3– Build your LinkedIn presence
→ Create a clean, professional profile and engage with people in your field.
→ Share your journey, post projects, and connect with UK professionals.
→ This will pay off big time, don’t start late!

4– Get familiar with UK culture
→ Open up to the world. Speak to people. Try different foods. Have fun.
→ You’ll connect faster and feel more confident in conversations and group projects.

5– Set your routine early
→ Bring back your home routine, things that keep you sane.
→ Balance studies with something that keeps you active or creative, sports, gym, etc…
→ It’s easier to stick to good habits than rebuild them after burnout.

—

Do these before you land and you’ll hit the ground running, academically, socially, and professionally.


PS: Full YouTube video in the comments below

Most data scientists don’t fail because they’re lazy.
They fail because they’re stuck learning instead of building.

Here’s why 👇

PS: Full video on YouTube, link in comments

1. 𝗧𝗵𝗲 𝗖𝗼𝗻𝘀𝘂𝗺𝗲𝗿 𝗘𝗳𝗳𝗲𝗰𝘁

You’re trapped in tutorial hell.
You binge courses, follow every trend, and do what everyone else is doing.
You feel busy, but your portfolio looks the same after months.

→ You’re consuming knowledge… not applying it.

2. 𝗧𝗵𝗲 𝗜𝗹𝗹𝘂𝘀𝗶𝗼𝗻 𝗼𝗳 𝗣𝗿𝗼𝗴𝗿𝗲𝘀𝘀

You think you’re improving, you’re not.
You’re collecting shallow skills with no depth.
A bit of Python, a bit of SQL, a bit of Power BI…
But nothing solid enough to make you stand out.

→ Learning feels like progress, until you try to build something real.

3. 𝗧𝗵𝗲 𝗕𝘂𝗶𝗹𝗱𝗲𝗿 𝗘𝗳𝗳𝗲𝗰𝘁

Real growth starts when you build.
Pick one complex project that scares you a little.

Learn what’s needed along the way.
Read the docs. Use ChatGPT. Open books.
Stick to a few tools. and master them before adding new ones.

→ That’s how you turn learning into experience.

---

Stop being a content consumer.
Start being a builder.

You don’t need 100 certificates. 
You need one project that changes how you think.

📌 YT video link:  https://lnkd.in/eegvYUdR

Watching tutorials doesn’t make you a data scientist

Let’s be honest:

→ Watching tutorials isn’t doing the thing.
→ Making a plan to learn Python isn’t doing the thing.
→ Talking about your next ML project isn’t doing the thing.
→ Writing “build portfolio” on your to-do list isn’t doing the thing.
→ Even feeling bad for not doing the thing… still isn’t doing the thing.

The only thing that’s doing the thing, is DOINT IT.

For data scientists, that means:

→ Hitting run even when you’re not sure it’ll work.
→ Writing that post to share what you learned.
→ Opening your notebook and coding.
→ Cleaning that messy dataset.
→ Shipping that model.

Stop preparing to do the thing.
Start doing the thing.

Ambition is closing the gap yourself.
Entitlement is expecting others to close it for you.

Your career grows when you act, not when you plan to act.

Post inspired by Chris Williamson on taking action.

If you’ve ever shared a Power BI report, read this.
10 RLS tips that every data analyst must know 👇

1️⃣ 𝗪𝗵𝗮𝘁 𝗥𝗟𝗦 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗱𝗼𝗲𝘀
It filters data per user so each person only sees what they should.
Example: a sales manager sees only their region’s sales, not everyone else’s.

2️⃣ 𝗗𝗲𝗳𝗶𝗻𝗲 𝗿𝗼𝗹𝗲𝘀 𝗶𝗻 𝗣𝗼𝘄𝗲𝗿 𝗕𝗜 𝗗𝗲𝘀𝗸𝘁𝗼𝗽
Go to Modeling → Manage Roles and add filters like:
[Region] = "UK"

→ Test it with “View as Role” before you publish.

3️⃣ 𝗔𝗹𝘄𝗮𝘆𝘀 𝘁𝗲𝘀𝘁 𝗯𝗲𝗳𝗼𝗿𝗲 𝗽𝘂𝗯𝗹𝗶𝘀𝗵𝗶𝗻𝗴
RLS doesn’t auto-apply in Desktop, you’ll always see all data by default.
✅ Use “View as Role” to check exactly what each user will see.

4️⃣ Make it dynamic
Avoid static filters, use the user’s email directly:
[Email] = USERPRINCIPALNAME()
This way, each logged-in user only sees their rows automatically.

5️⃣ Use a mapping table for scalability
For large setups, build a user-region mapping table and filter dynamically with USERPRINCIPALNAME())

6️⃣ Why it doesn’t “work” in Desktop
By default, Desktop shows all data because there’s no real login context.
You must test with “View as Role”, that’s the only way to simulate RLS locally.

7️⃣ RLS can be bypassed by permissions
If someone has Build, Edit, or Admin rights, RLS won’t stop them.
Keep those roles to a minimum, use Viewer or App access for restricted users.

8️⃣ Be careful with composite models
If your dataset mixes Import + DirectQuery, RLS may not pass through properly.
Keep RLS on the main model or simplify relationships where possible.

9️⃣ USERPRINCIPALNAME() vs USERNAME()
They look similar but aren’t:
USERPRINCIPALNAME() → returns email (e.g. alex@company.com)
USERNAME() → returns domain\user (e.g. COMPANY\alex)
→ Use the one that matches your organisation’s identity setup.

🔟 Combine multiple filters safely
When you filter by several fields (region, department, etc.), use AND carefully
One mismatch can hide all rows; always test combinations.


RLS is simple in theory but tricky in practice.
Most issues come from missing relationships, wrong functions, or permission levels.

What’s the most frustrating part of RLS you’ve faced in Power BI?

📌 Save this for your next project.
♻️ Share this to help your network in data.

taplio