Level up Your ML Game ❤️️| Writing a Lot | Data Mentor

DMs might get missed. For collaborations, please write to me at: cornelliusyudhawijaya@nb-data.com

Hi there👋

🔹 I help organizations turn data into scale products, bridging my background as a Data Scientist and AI Engineer with my current role as Chief Product Officer. I focus on data Product Strategy, transforming raw data into trusted, usable, and monetizable products.

🔹 With 7+ years leading enterprise data science teams (Allianz, startups) and now building SaaS at ARIF, I’ve seen how models, dashboards, and pipelines fail without a product mindset. That’s why I work on frameworks that move teams from data as an asset → data as a product.

Some highlights from my journey:
✅ Propensity models that drove 150% of Allianz’s annual revenue targets
✅ Fraud detection systems reducing losses by 7-figure margins
✅ LLM-powered chatbots with RAG cutting customer support costs by 30%
✅ SaaS data platform (ARIF) turning advanced analytics into a productized experience

🔹 I also share what I learn. I’ve written 500+ articles (KDnuggets, Towards Data Science, MachineLearningMastery, Non-Brand Data) for 2M+ monthly readers, translating complex AI/ML + data strategy concepts into actionable playbooks.

🔹 Today, I collaborate with professionals and organizations on:
👉 Consulting: Data product strategy, machine learning pipelines, and LLM product design
👉 Content & Speaking: Articles, workshops, and keynotes on Data Science and AI

📌 If you want to bridge the gap between data teams and business outcomes. Let’s talk.
📩 Write to me for consulting, speaking, or content partnerships.

Cornellius Yudha Wijaya

Open to Collaborate | Storyteller | Lead Data Engineer@Wavicle| Linkedin Top Voice 2024 | Globant | Linkedin Learning Instructor | 2xGCP & AWS Certified | LICAP'2022

Data Engineer with 7+ years of experience working on data engineering, analysis, and modelling projects for various domains, such as banking, healthcare, sales, and e-commerce. I leverage my strong statistical and analytical ability to explore business ideas and provide innovative solutions. 

Skilled with technologies and strong intellect with enthusiasm as an avid learner with challenging opportunities. I’m working as a LinkedIn Learning Instructor to bring change as a data evangelist.

Post Graduated in Master of Technology in Data Science and Engineering from Birla Institute of Technology and Science, Pilani in 2023, where I learned advanced concepts and techniques in data science and engineering, machine learning, cloud computing, and visualization. 

I hold multiple certifications in GCP, AWS, and Azure, and I enjoy working with cloud technologies and big data platforms.

I am passionate about data science and women empowerment, and I have been recognized as the 
- Linkedin Top Voice 2025, 2024
- Data Science and Women Empowerment Thought Leader by Global AI Hub. 

I am also a proud participant of the LinkedIn Creator Accelerator Program 2022, where I share my insights and experiences with the data science community. 

I am open to collaborate and share ideas, or explore various opportunities in the data field.

Pooja Jain

The Great Orange Processing Wars: ETL vs ELT vs EtLT

Three factories. Three philosophies. 
One truth most data engineers miss.

𝗘𝗧𝗟 — The Juice Bar
Squeeze before you store. Quality-first, schema-first, compliance-first.

Works perfectly until your business logic changes — then you re-run everything.
Best for: fraud detection, classical ML, regulated industries.
Hidden cost: you can't un-squeeze juice. Raw fidelity is gone forever.
These need strict compliance and clean, pre-processed data before loading.

𝗘𝗟𝗧 — The Warehouse
Store everything raw. Transform strictly on demand. Pure flexibility.

Works perfectly until nobody knows what's in Aisle 47 anymore.
Best for: ad-hoc analytics, LLM training data, fast-moving startups.
Hidden cost: "we'll clean it later" becomes never. Swamps, not lakes.
Here, raw data is loaded first, then transformed as needed—great for massive, unstructured, fast-changing datasets.

 𝗘𝘁𝗟𝗧 — The Gourmet Factory
The lowercase t changes everything.

E — Extract faithfully
t —  Fix only what must be fixed early (PII masking, dedup, type casting)
L — Load into columnar storage
T — Transform richly at query time, per use case

The small t handles what belongs at ingestion: compliance, deduplication, partitioning.
The big T handles what belongs at consumption: business logic, ML features, aggregations.
Best for: RAG pipelines, LLM training, GDPR systems, real-time personalization.
Used, when different data types need different strategies.

The real choice isn’t just technical; it’s about timing.
→ ETL commits early.
→ ELT commits late.
→ EtLT commits minimally early — but maximally late.

That tradeoff shows up everywhere in system design.
Eager vs lazy. 
Compiled vs interpreted. 
Normalized vs denormalized

💡 The Modern Data Engineering Playbook
Stop asking: "ETL or ELT?"
Instead, start asking: "What fuels my AI use case?"
🍊 Traditional ML? → Serve up that perfectly squeezed juice (ETL)
🥤 Exploratory AI? → Keep the whole fruits on hand, blend them later (ELT)
⚡ AI at Scale? → Prepare smartly, create a variety (EtLT)

🎯 The Truth Bomb
AI is like the promised land—bringing automation, intelligence, and scalability.
But without the right pipeline? It’s just chaos.
The best data engineers don’t play favorites.
They choose the right tool for the right need.

What’s your philosophy on data pipelines? 👇
Share it in the comments!

🔍 𝗔–𝗭 𝗼𝗳 𝗗𝗮𝘁𝗮 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴: 𝗬𝗼𝘂𝗿 𝗨𝗹𝘁𝗶𝗺𝗮𝘁𝗲 𝗞𝗻𝗼𝘄𝗹𝗲𝗱𝗴𝗲 𝗠𝗮𝗽!

In a world of limited time but unlimited tools, 
Data Engineers are expected to master everything — from ingestion to orchestration, from storage to serving, and from DevOps to AI.

The demand is booming. 
The ecosystem is evolving.
And the lifecycle? 
It’s massive.

📚 Sharing the A–Z snapshot of the tools & frameworks every data engineer should know:-
 
 𝗔 – Airflow, Avro, AWS
 𝗕 – BigQuery, Beam, Bigtable
 𝗖 – Cassandra, Cloudera, Couchbase
 𝗗 – Databricks, dbt, Dagster
 𝗘 – Elasticsearch
 𝗙 – Flink, Fivetran, Fault Tolerance
 𝗚 – Git, Glue, Grafana
 𝗛 – Hadoop, Hive, Hudi
 𝗜 – Iceberg, InfluxDB
 𝗝 – Jenkins, Jupyter, JSON
 𝗞 – Kafka, Kubernetes, Kestra
 𝗟 – Looker, Lakehouse, Luigi
 𝗠 – MongoDB, Metabase, Matillion
 𝗡 – NiFi, Neo4j
 𝗢 – OLTP/OLAP, Observability
 𝗣  – Python, PostgreSQL, Prefect
 𝗤 – Query Optimization, Qlikview
 𝗥 – Redshift, RabbitMQ, Redis
 𝗦 – Spark, Snowflake, SQL
 𝗧 – Terraform, Tableau, Talend
 𝗨 – Unix, Unity Catalog
 𝗩 – Versioning, Vertica
 𝗪 – Workflow Management, Warehousing
 𝗫 – XML, XGBoost
 𝗬 – YAML, YARN
 𝗭 – Z-Ordering, ZooKeeper

📍 Bonus: Don’t miss emerging trends like Data Mesh, LLMs, and Generative AI reshaping the future of data engineering.

👉 What tools or concepts would you add to this list?
Drop your favorite in the comments and let’s build the most complete A–Z cheat sheet together!

#Data #Engineering #ETL #BigData #Cloud #AI

No alternative text description for this image

The Iceberg nobody shows you about Data Engineering!

Data engineers don't surface level things like,
Spark and Kafka and Snowflake
$150k salary dreams
Building ML pipelines
Modern data stack vibes

Here’s how the life of a ℝ𝔼𝔸𝕃 𝔻𝔸𝕋𝔸 𝔼ℕ𝔾𝕀ℕ𝔼𝔼ℝ looks like  😮‍💨
(The brutal truth nobody tells you)

1️⃣ 𝗖𝗮𝗿𝗲𝗲𝗿 𝗖𝗵𝗮𝗻𝗴𝗲𝗿: “I’ll master Spark, Kafka, and Snowflake in 3 months and land a $150K job!”
2️⃣ 𝗝𝗼𝗯 𝗛𝘂𝗻𝘁𝗲𝗿: “Why do they want 5 years experience for an ‘entry-level’ position? And what’s a ‘take-home coding challenge’?”
3️⃣ 𝗜𝗻𝘁𝗲𝗿𝘃𝗶𝗲𝘄 𝗖𝗮𝗻𝗱𝗶𝗱𝗮𝘁𝗲: “I spent 40 hours building their pipeline assignment. They ghosted me after the final round.”
4️⃣ 𝗝𝘂𝗻𝗶𝗼𝗿 𝗗𝗮𝘁𝗮 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿: “The production data is nothing like the tutorial datasets. Why is everything broken and undocumented?”
5️⃣ 𝗗𝗮𝘁𝗮 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿: “I’m not building ML models. I’m fixing CSV parsing errors and explaining why dashboards show different numbers.”
6️⃣ 𝗦𝗲𝗻𝗶𝗼𝗿 𝗗𝗮𝘁𝗮 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿: “60% of my time is meetings about data governance. The other 40% is arguing with stakeholders about SLAs.”
7️⃣ 𝗟𝗲𝗮𝗱 𝗗𝗮𝘁𝗮 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿: “The pipeline that worked for 2 years suddenly failed. Nobody knows why. The person who built it left 6 months ago.”
8️⃣ 𝗦𝘁𝗮𝗳𝗳 𝗗𝗮𝘁𝗮 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿: “I spend more time in Slack explaining why data is 2 hours late than actually engineering. Also, our AWS bill is $50K this month.”
9️⃣ 𝗗𝗮𝘁𝗮 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁: “I designed the perfect architecture. Nobody followed it. Everything runs on Excel sheets shared via email. I’ve accepted this.”
🔟 𝗖𝗧𝗢 (𝗳𝗼𝗿𝗺𝗲𝗿 𝗗𝗮𝘁𝗮 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿): “We need to be more data-driven. Also, can you build this by Friday? And make it real-time. Budget is $500.”

The real job description: 
• 20% building cool stuff 
• 80% debugging, meetings, and existential crisis

What they don’t tell you:
→ You’ll become a CSV therapist
→ “Data quality” means fixing the same broken pipeline every week
→ The hardest part isn’t the tech—it’s managing expectations
→ It's about business chaos and technical constraints
→ Data quality isn't a tool. It's an ownership issue
→ Most failures aren't technical. They're communicational
→ Scalability is easy to achieve. Designing and enforcing are two different things
→ The more senior you get, the less you're involved

The most challenging production bug is about stakeholder misalignment.
"Real-time" data usually means we didn't plan for this. 

Data engineers, put differently, are part therapists, detectives, and negotiators.

And despite the 2AM alerts, data pipeline issues and never ending slack threads —

It’s a privilege because in the end, 
the pipeline runs, 
the dashboard matches,
… And leadership finally trusts the numbers. Pure magic ✨

But here’s the thing… Despite all this, we still love what we do.

Do you remember, when did you automate your first pipeline?

Dear Data Engineers — meet your new coworker. It's not human.
Monday, 9:47 AM. Analyst pings: "orders_final is missing 10K rows."

You: Open Airflow. Query source. Compare counts. Find schema drift. Manual backfill. 45 minutes. Annoyed.
The Agent: Queries metadata → checks logs → detects schema drift → generates fix → asks "Run this?" 90 seconds. Done.

Here's the analogy that makes it click:

AI Agents = Self-modifying Airflow DAGs
𝖸𝗈𝗎 𝗁𝖺𝗋𝖽𝖼𝗈𝖽𝖾: 𝖾𝗑𝗍𝗋𝖺𝖼𝗍() → 𝗍𝗋𝖺𝗇𝗌𝖿𝗈𝗋𝗆() → 𝗅𝗈𝖺𝖽()

Agents do:
𝗐𝗁𝗂𝗅𝖾 𝗇𝗈𝗍 𝗌𝗈𝗅𝗏𝖾𝖽:
  𝗍𝗁𝗂𝗇𝗄() → 𝗉𝗂𝖼𝗄_𝗍𝗈𝗈𝗅() → 𝖾𝗑𝖾𝖼𝗎𝗍𝖾() → 𝖾𝗏𝖺𝗅𝗎𝖺𝗍𝖾()

Your DAG just became sentient and stopped asking for help.

The good news? You already have 80% of the skills.

Your Python scripts ARE the tools. You just need to make them agent-callable — add clear docstrings, wrap them as functions, expose them to the loop. 

The real shift isn't technical. It's mental.

Before: you were the pipeline. 
After: you design the system that runs the pipeline.
Before: 2 AM Slack pings wake you up. 
After: the agent handles it. You review the decision over morning coffee.

You're not being replaced. You're being scaled.
One DE + three agents = a five-person team's output.
Agents handle 2 AM fires. You architect system!

Working with AI Agents? Here are a few rules I always find myself returning to.

Do's:
✅ Debug it weekly? Automate it. Agents thrive on tools, not just shared knowledge.
✅ Document as if the reader knows nothing. Trust me, the agent definitely doesn’t.
✅ Record every decision made by agents. When things go wrong, you’ll want a clear trail.

Don'ts:
❌ Granting write access on Day 1. Start with read-only. Trust is something you build over time.
❌ Skipping human checkpoints on critical operations. Always gate anything that changes data.
❌ Assuming agents understand your context. They don’t. Make it crystal clear.

What’s the takeaway?
Agents reflect your habits—both the good and the bad.
So, build cleanly. Document thoroughly. Keep communication open where it counts.

You were building systems for humans,
Now build them for Agents.
This post is inspired by Vu Trinh's article on AI Agents. Find the link below. 👇

🌊 A fish thinks water needs it. Water thinks it needs the fish.

Then the net drops.

The fish gets captured. 
The water? It simply flows through.

𝗧𝗵𝗶𝘀 𝗶𝘀 𝗲𝘅𝗮𝗰𝘁𝗹𝘆 𝘄𝗵𝗮𝘁’𝘀 𝗵𝗮𝗽𝗽𝗲𝗻𝗶𝗻𝗴 𝗶𝗻 𝗔𝗜 𝘁𝗼𝗱𝗮𝘆.

Raw data thinks AI can't survive without it. 
AI systems think any data is worth training on.

Then comes the real test — production.

And suddenly:

🐟 Ungovernated data gets 𝘁𝗿𝗮𝗽𝗽𝗲𝗱 in compliance failures
🐟 Unobserved pipelines get 𝗰𝗮𝘂𝗴𝗵𝘁 in silent corruption
🐟 Poor quality features get 𝗻𝗲𝘁𝘁𝗲𝗱 by model drift
But the water — the Governance framework — just flows right through.

It was never enslaved by bad data.

Here's the truth nobody tells you:
Most organizations treat data like fish — abundant, assumed, and disposable.

The ones building real AI treat data like whales.

Whales can't be netted…
They're too governed, too observable, too quality-conscious to be trapped by broken systems.

The 3 pillars that separate fish from whales:

⚙️ 𝗗𝗮𝘁𝗮 𝗚𝗼𝘃𝗲𝗿𝗻𝗮𝗻𝗰𝗲 = The ocean itself (Ownership, lineage, policy — the environment data lives in)

👁️ 𝗗𝗮𝘁𝗮 𝗢𝗯𝘀𝗲𝗿𝘃𝗮𝗯𝗶𝗹𝗶𝘁𝘆 = Knowing where the whale is (Freshness, volume, schema, distribution — real-time awareness)

✅ 𝗗𝗮𝘁𝗮 𝗤𝘂𝗮𝗹𝗶𝘁𝘆 = The whale's immune system (Completeness, accuracy, consistency — what makes it survive production)

The ocean doesn’t mourn the fish lost to the net.
It only protects what chose to become sovereign.
Your data can survive capture — or it can command the ocean.

𝐴𝐼 𝑑𝑜𝑒𝑠𝑛’𝑡 𝑓𝑎𝑖𝑙 𝑏𝑒𝑐𝑎𝑢𝑠𝑒 𝑜𝑓 𝑏𝑎𝑑 𝑚𝑜𝑑𝑒𝑙𝑠. 𝐴𝐼 𝑓𝑎𝑖𝑙𝑠 𝑏𝑒𝑐𝑎𝑢𝑠𝑒 𝑜𝑟𝑔𝑎𝑛𝑖𝑧𝑎𝑡𝑖𝑜𝑛𝑠 𝑓𝑖𝑠ℎ 𝑓𝑜𝑟 𝑑𝑎𝑡𝑎 𝑖𝑛𝑠𝑡𝑒𝑎𝑑 𝑜𝑓 𝑟𝑎𝑖𝑠𝑖𝑛𝑔 𝑤ℎ𝑎𝑙𝑒𝑠.

💬 Are you fishing for data or raising whales?
#AI #Data

Communication is the key - From raw data to profitable outcomes!

It’s the TRUST,
With CLEAN outcomes,
To deliver solutions with IMPACT!

…. That helps you refine — Raw data into reliable solutions to deliver PROFIT.

Most data and AI teams are sitting on a treasure trove — and they might not even realize it.  

What separates a $1M budget line from a $10M+ profit powerhouse? 
It’s not just your algorithm or your tech stack. It’s all about how you navigate through these five stages:  

1️⃣ RAW DATA = The Ore It’s messy, unstructured, and often trapped in databases that no one fully trusts. Every company has it, but very few know how to refine it.  

2️⃣ SORTED = Refined Ingots Now we’re talking! Cleaned, labeled, and structured — think Metal, Copper, Gold, Silver. It’s got form, but just having form doesn’t mean it’s market-ready.  

3️⃣ ENGINEERED AI = The Component This is where your machine learning models, logic layers, and automation triggers come into play. Your engineers have crafted some impressive parts, but they’re still just that — parts. 

4️⃣ INTEGRATED SOLUTION = The Product Costs drop ↓ and revenues rise ↑. Now you’ve got something the market can actually USE. This is where the $10M+ potential lies.  

5️⃣ EXPLAINED STORY = The ROI “We transformed from a $1M cost center into a $10M profit machine.” This is the kind of statement that gets budgets approved, deals sealed, and careers launched.  

Here’s the hard truth:  
Most engineers stop at Stage 3, 
while many managers only get to Stage 4. 

The true winners — the leaders who really excel — are the ones who thrive at Stage 5.  

Keep this in mind:
Communication transforms beans into drinks.
Trust transforms drinks into habits.
Impact transforms habits into profits.

Stop building houses. 
Start constructing Profit Factories.  
That’s the 10X leap.  

♻️ Save this if you’re leading a #data or #AI team. 
💬 Which stage is your team currently stuck in? Let us know in the comments!

When DATA GOVERNANCE comes as an afterthought, you're literally screwed!

Agree?
Ignoring foundations that builds data trust can be painful.

Someone once asked me — Pooja, do you really think a simple data pipeline could cost such a huge bill?

Absolutely, it can. 
Data solution that is easy to represent.
But, here's the common pitfall many teams encounter:

→ The pipeline looks pristine on a whiteboard
→ It’s easy to draw. Easy to demo.
→ But scaling it? That’s a whole different ballgame. Trusting it? Even harder.

Here's what nobody tells you right off the systems:

📌 A sleek diagram ≠ a dependable data system
📌 Quick data movement ≠ accurate data movement
📌 More pipelines without governance = more confusion, not more clarity

What does DATA GOVERNANCE truly offers the teams?

✅ Everyone knows "who’s responsible" for what data
✅ No more debates over "which version is the right one?"
✅ Teams spend less time troubleshooting — and more time making "decisions"
✅ Your data builds "trust". Both internally and externally.

What’s the real price of skipping it?

It’s not just about the money.
It’s about team frustration.
It’s about flawed decisions.
And it’s about data that no one trusts.

Governance isn’t just a tech problem.
It’s a people + process + discipline problem.

Build the foundations before you start building floors.

Start small, but make it intentional.
Be honest, don't let a chat with the CFO at 9 AM cause chaos.
The $1150K surprise bill on a Monday morning can be thrilling over the week 😬

Here’s are some amazing folks to follow and learn more on Data and AI governance —

What's your perspective data leaders? 
George Firican 
John Wernfeldt 
Clare Kitching 
Charlotte Ledoux  
Dylan Anderson  
Piotr Czarnas 

What’s the one governance gap your team keeps revisiting?
Share it below 👇

taplio

The Battle

Cornellius Yudha Wijaya

Pooja Jain

Followers

Posts

Engagements

Likes

Performance ScorePooja Jain's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Cornellius Yudha Wijaya

Pooja Jain

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);}Pooja Jain's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScorePooja Jain's LinkedIn Performance