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

Two months. 250M views. 700K comments.

One question: who should Steven Bartlett invite next?

I’ve always loved The Diary of a CEO.

It’s one of those podcasts that makes you think deeper, 
about business, people, and purpose.

So I decided to turn my curiosity into a real AI project.

For the past two months (on the side of my main work), I built an AI system that analysed every episode of The Diary of a CEO in 2025, every topic, every guest, every comment.

Here’s what it did:

- Ran 𝘁𝘄𝗼 𝘁𝘆𝗽𝗲𝘀 𝗼𝗳 𝗮𝗻𝗮𝗹𝘆𝘁𝗶𝗰𝘀:
  - 𝗧𝗼𝗽-𝗳𝘂𝗻𝗻𝗲𝗹: measuring episode performance through views, likes, and engagement metrics. Given them different weights.
  - 𝗖𝗼𝗺𝗺𝗲𝗻𝘁 𝗮𝗻𝗮𝗹𝘆𝘀𝗶𝘀: using NLP, LLMs and SLMs to understand audience sentiment at scale
- Combined both layers to see what drives real engagement beyond the surface

This project wasn’t just about insights; it was about "learning".

I wanted to:

- Build a full 𝗲𝗻𝗱-𝘁𝗼-𝗲𝗻𝗱 𝗔𝗜 𝘀𝘆𝘀𝘁𝗲𝗺 from scratch
- Handle real, messy text data at scale (emojis, sarcasm…)
- Use AWS GPU servers for training and Streamlit for deployment
- Actually ship something meaningful

It reminded me why I love data.

You don’t need to do random tutorials or fake datasets.

You can take something you’re genuinely passionate about, podcasts, football, fitness and turn it into a real AI project that teaches you 10x more.

𝗧𝗵𝗶𝘀 𝗼𝗻𝗲 𝘁𝗼𝗼𝗸:

💻 Python, NLP, LLMs, SLMs
☁️ AWS for compute
🕓 2 months (nights + weekends)
💰 ~$100 total cost

Based on the analytics, the top 5 topics were 𝗧𝗲𝗰𝗵, 𝗣𝗼𝗹𝗶𝘁𝗶𝗰𝘀, 𝗥𝗲𝗹𝗶𝗴𝗶𝗼𝗻, 𝗣𝗲𝗿𝘀𝗼𝗻𝗮𝗹 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗺𝗲𝗻𝘁, 𝗮𝗻𝗱 𝗙𝗶𝗻𝗮𝗻𝗰𝗲,

And the most requested guest by far was 𝗘𝗹𝗼𝗻 𝗠𝘂𝘀𝗸.

Who knows… 

With more time and resources, I could make this even better.

Comment below if you want me to make a detailed video on how I built this.

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

XGBoost is (still) one of the best ML models in practice.

Here’s why:

If you work with structured/tabular data, there’s a high chance XGBoost will beat most other models, even deep learning in many cases.

But what actually makes it so good?

𝗪𝗵𝗮𝘁 𝗶𝘀 𝗫𝗚𝗕𝗼𝗼𝘀𝘁?

XGBoost stands for Extreme Gradient Boosting.

It builds a group of weak learners (decision trees), each one fixing the mistakes of the previous ones.

It’s part of the gradient boosting family, but far more optimized for speed and memory.

== == == 

𝗪𝗵𝘆 𝗶𝘁 𝘄𝗼𝗿𝗸𝘀 𝘀𝗼 𝘄𝗲𝗹𝗹:

- Handles missing values on its own
- Built-in regularization helps avoid overfitting
- Parallel training makes it strong on large datasets
- Tree pruning and early stopping keep the model from getting too complex
- Works for classification, regression, ranking, and can be used for time-series with feature engineering

== == == 

𝗧𝗿𝗮𝗱𝗲𝗼𝗳𝗳𝘀:

- Can train slower than random forests and simple models
- Can overfit on small datasets if not tuned
- Not as clear to read as linear models
- LightGBM is sometimes faster on huge datasets, but results are often similar

== == == 

𝗞𝗲𝘆 𝗛𝘆𝗽𝗲𝗿𝗽𝗮𝗿𝗮𝗺𝗲𝘁𝗲𝗿𝘀 𝘁𝗼 𝘁𝘂𝗻𝗲:

- n_estimators → number of trees
- max_depth → how deep each tree can grow
- learning_rate → how strong each new tree is
- subsample, colsample_bytree → add randomness for stability
- gamma, lambda, alpha → regularization
- (Optional but strong) min_child_weight → helps avoid overfitting

== == == 

𝗪𝗵𝗲𝗻 𝘁𝗼 𝘂𝘀𝗲 𝗶𝘁:

- You’re working with cleaned tabular data
- You care about accuracy more than full explainability
- You want strong results without complex deep learning setups
- You don’t have GPU power for heavy neural nets
- You need a strong baseline for both competitions and real work

XGBoost is not flashy.
But it’s reliable, fast, and gets the job done when tuned well.

📌 Save this post for your next ML project.

95% of data scientists who reach out to me

Have this 1 thing in common ⤵️ 

“They excel at building ML models in a notebook”
but "They struggle to build something beyond a Demo"

Look, I’m not against notebooks, but that mindset is restrictive.

If your project only:

→ runs on your machine
→ uses static data (all clean)
→ isn’t deployed and usable

Then it’s a "SIMPLE" demo, not a "REAL" data project.

In my recent journey to bridge the gap, I’ve decided to collaborate with Markus Kuehnle on a detailed carousel that includes major points of an end-to-end ML project, such as:

→ Setting project environment
→ Exploratory data analysis
→ Feature Engineering
→ Model Registry & Experiment Tracking
→ Feature, training, inference pipelines
→ Serving layer (FastAPI)
→ Docker container (reproducibility)
→ GitHub Actions (CI/CD)
→ Deployment & monitoring

Hope it gives you more clarity on what it takes to build and deploy an ML model end to end.

PS: Find hands-on tutorials on this on my YouTube channel

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

Machine Learning End-to-End pipeline

The number one thing that 10x’d my life.
Personal and professional combined.

It’s sending a short recorded video with Loom.

→ When applying for jobs, adding a short video makes it personal.
→ When doing outreach, a quick recording shows you did your research.
→ When a colleague needs help but can’t meet, a short video solves it.
→ When a friend asks for feedback on a post, I send a quick video.
→ When working on a SaaS or project, a weekly recording keeps progress visible.

It builds trust, saves time, and makes communication human again.

What’s your main cheat code to improve work or relationships in the digital world?

Before: MVP in 90 days.

Now: MVP in 6 hours.

There’s no excuse not to build anymore, 
not with 𝗥𝗲𝗽𝗹𝗶𝘁.

Whether it’s a business idea, a personal project, or a side hustle, you can now bring it to life using plain English.

No setup. No API keys. No coding skills needed.

Why Replit stands out

* Build in plain English.
* Vibe code anywhere.
* All-in-one workspace.
* AI integrations built-in. 
* Connect your tools.

Creators are already using Replit to:

* Launch SaaS ideas over a weekend
* Build AI-powered dashboards
* Automate boring workflows
* Turn sketches into real, working apps

I recently used it to build a data analytics scoping app, something that used to take days, done in under an hour.

Replit makes it possible to build ideas, not just talk about them.

Try it for yourself → https://replit.com

PS: Tag me if you build something cool, I’d love to see what you create.

🐘 Get 4 - 5+ warm Leads/week || I help founders through LinkedIn Lead Generation through content + DMs || Reach out for Lead Gen + Personal Branding services

Struggling with repetitive tasks and scaling your business?

If you're a Small business owner. 

I help you implement AI-powered automation solutions to:

1. Save time by automating repetitive tasks.
2. Generate leads effortlessly with workflows.
3. Improve customer experience with AI chatbots.

What I Do:

1. Automate lead generation, emails, and workflows.
2. Set up chatbots for 24/7 customer support.
3. Optimize operations to save you 10+ hours/week.

Offers:

1. Done-for-You Automations: 

Customized systems for lead gen, 
task automation, and 
data management.

2. 1:1 Consulting: Learn No-code tools like Make to automate your business in 90 days.

---

Want to scale effortlessly?

📧 Email me at: Kavinprasathofficial@gmail.com

DM me “AI Solutions” to get started today!

📌 PS: You can wait and keep struggling—or take action and automate your success now.

Kavin Prasath 🐘

Most D2C brands hit a ceiling—not from lack of demand, but from operational drag. 

• Manual campaigns, 
• one-size-fits-all emails, and 
• disconnected customer journeys slow growth and erode retention. 

The edge now belongs to brands that automate smarter and scale faster.

Marketing automation is no longer optional infrastructure—it’s how modern D2C funnels actually work. 

• From abandoned carts to post-purchase follow-ups, 
• automation connects the dots across CRM, 
• customer behavior, and 
• channel performance. 

But the impact is deeper than just efficiency.

Here are three core reasons automation is now essential for D2C brands:

1️⃣ Cart recovery needs split-second timing. 
• The average D2C cart abandonment rate is over 65%. 
• Manual campaigns miss the critical first hour when intent is still high. 
• Automated flows using triggers like browse behavior and cart value can recover 10–15% of lost revenue—without lifting a finger after setup.

2️⃣ Customer journeys aren’t linear. 
• Most shoppers don’t move from ad to checkout in one clean path. 
• Effective funnels require dynamic nurturing—email flows, SMS touchpoints, product reminders—that adapt to behavior. 
• Automation lets brands design CRM systems that respond in real time, guiding users from discovery to conversion without manual effort.

3️⃣ Retention starts with relevance. 
• Repeat customers drive profitability, but generic campaigns don’t build loyalty. 
• Automated segmentation allows post-purchase flows, replenishment reminders, and tailored offers at scale—turning one-time buyers into regulars through timely, personalized communication.

In short: automation turns reactive marketing into a proactive growth engine. It doesn’t just save time—it orchestrates every touchpoint with purpose, context, and speed.

For D2C teams building sustainable growth, automation isn’t a quick fix. It’s the foundation.

📌 𝗣𝗦: 𝗙𝗼𝗹𝗹𝗼𝘄 👉 𝗜'𝗺 Kavin Prasath 🐘, 𝗣𝗼𝘀𝘁𝗶𝗻𝗴 𝗱𝗮𝗶𝗹𝘆 𝗮𝗯𝗼𝘂𝘁 𝗔𝗜 𝗳𝗼𝗿 𝗕𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗼𝘄𝗻𝗲𝗿𝘀

📌 𝗪𝗮𝗻𝘁 𝘁𝗼 𝗯𝘂𝗶𝗹𝗱 𝗔𝗜 𝗮𝗴𝗲𝗻𝘁𝘀 𝗳𝗼𝗿 𝘆𝗼𝘂𝗿 𝗯𝘂𝘀𝗶𝗻𝗲𝘀𝘀? 𝗟𝗲𝘁’𝘀 𝗰𝗼𝗻𝗻𝗲𝗰𝘁 𝗮𝗻𝗱 𝗺𝗮𝗸𝗲 𝗶𝘁 𝗵𝗮𝗽𝗽𝗲𝗻! 🫂

#ai #partner #owner #founder #leader

👉 | BDM at INovaBeing 👉 | AI Transformation Partner 👉 | We build AI Multi Agents & Automation for businesses 📍 DM Me to know more 👇

taplio

The Battle

Anas Riad

Kavin Prasath 🐘

Followers

Posts

Engagements

Likes

Performance ScoreAnas Riad's LinkedIn Performance

Performance ScoreKavin Prasath 🐘's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Anas Riad

Kavin Prasath 🐘

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Anas Riad's LinkedIn Performance

Performance ScoreKavin Prasath 🐘's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreAnas Riad's LinkedIn Performance