NannyML

NannyML is an Open Source Python library to detect silent model failure, estimate the performance of ML model even without ground truth, and robustly identify data drift.

Co-Founder and CEO

Prophecy Labs

Founder

Data Solutions. Elegantly Crafted.

Prophecy Labs helps companies automate complex work, optimize business processes, and predict events that impact your business.

Let our passion, be your edge.

Managing Partner

Essent.be

Developing a data platform in Scala/Spark to deal with large scale data and
deploying large scale machine learning models.

Big Data Engineer

SCK-CEN

This project was to help prepare for long term space travel namely a long term lunar outpost within the following decade. It involved working with gene expression data to help determine the effects of long term isolation and harsh condition on the human immune system. Data from workers of the Concordia platform in antarctica was used. Technical details of the R code can be found at the link below.

Research Bioinformatician 

Building an open source library to estimate the performance of deployed machine learning models in the absence of ground truth.

I love talking about: machine learning, decision making, post deployment data science, performance estimation, and bunch of other stuff. 

currently learning about: causal inference 

Always open to have a chat 🙂

Leave me anonymous feedback -> https://www.admonymous.co/hakim_elakhrass

🥸Hakim Elakhrass

building Soda to monitor the worlds data and decisions

Doctoral Researcher (CS) at University of Helsinki

Last week at the NeurIPS 2025 presenting our paper "Estimating Model Performance Under Covariate Shift Without Labels", which is joint work with Jakub Białek, Wojtek Kuberski, and Nikolaos Perrakis. What a highlight reel of talks, presentations and discussions! For those who can't wait until the proceedings are published, check out the paper in arXiv at https://lnkd.in/dME4FhMT.

In2IT Technologies

- As a Business Analyst, conducted research, analyzed business operations, and solved the business 
  problems
- Collaborating with cross-functional teams, including product managers,  and other 
   stakeholders, to gather requirements, analyze data, and define the product roadmap
- Preparing various functional specification documents, facilitating review of the same by development 
   team
- Created detailed user stories, workflows, and acceptance criteria to ensure clear communication of business needs to development teams.
- Currently working on a Dashboard Product for a client.

Business Analyst

Octaculus

- Supported seniors in gathering and documenting requirements for product development initiatives.
- Developed process flow diagrams and use case documentation, streamlining internal processes and 
   reducing redundancies.
- Analyzed data trends and user feedback, providing actionable insights that improved product usability 
   and customer satisfaction.
- Assisted in sprint planning and backlog grooming sessions, maintaining alignment with Agile principles.
- Analyzing data for generating marketing strategies

Associate Business Analyst

DataDice Technologies

- Collect, clean, and preprocess structured and unstructured datasets.
- Conduct exploratory data analysis (EDA) to uncover trends, patterns, and insights.
- Assist in building dashboards and visualizations using tools like Power BI.
- Collaborate with cross-functional teams to understand business requirements and translate them into 
    analytical solutions.
- Support the team in automating data workflows and creating reports.
- Document findings and prepare presentations for stakeholders.

Data Analyst

Coaching Institute

- Provided one-on-one & group tutoring sessions in Social Studies(SST) & Mathematics, helping students improve their understanding & performance

-Assisted students in exam preparation, homework, & conceptual learning, leading to improved grades & confidence

- Focused on interactive learning methods, making subjects more engaging & easier to grasp

SST Tutor

• 𝐋𝐚𝐧𝐠𝐮𝐚𝐠𝐞𝐬: SQL
• 𝐃𝐚𝐭𝐚 𝐕𝐢𝐬𝐮𝐚𝐥𝐢𝐳𝐚𝐭𝐢𝐨𝐧: Power BI, Seaborn, Plotly
• 𝐅𝐫𝐚𝐦𝐞𝐰𝐨𝐫𝐤𝐬 & 𝐋𝐢𝐛𝐫𝐚𝐫𝐢𝐞𝐬: Streamlit, Pandas, NumPy, Matplotlib, Plotly, Scikit-learn.

Sachin Kumar

Most SQL beginners think DELETE, DROP, and TRUNCATE mean the same… until they break a production table.

1) DELETE:
is a Data Manipulation Language (DML) command used to remove records from a table.
It deletes specific rows based on conditions specified in WHERE clause.
DELETE does not alter the table structure; it only affects the data.
Use Case: Use DELETE when you want to selectively remove specific rows while keeping the table intact.


2) TRUNCATE:
is also a DDL command, but it focuses on removing data while preserving table structure.
It quickly removes all rows from the table without logging individual deletions.
Use Case: Use TRUNCATE when you want to clear out data efficiently without dropping the table.

3) DROP:
is a Data Definition Language (DDL) command that removes an entire table from the database.
It deletes both the table structure and all the data within it.
Irreversible: Once you DROP a table, it cannot be rolled back.
Use Case: Use DROP when you need to completely eliminate a table.


Day 68 : https://lnkd.in/gSUgzey9

#sql #database #dataanalyst #businessanalyst #productanalyst

𝐄𝐯𝐞𝐧𝐭 𝐥𝐨𝐠𝐬 𝐚𝐫𝐞 𝐧𝐨𝐢𝐬𝐲. 𝐒𝐐𝐋 𝐦𝐚𝐤𝐞𝐬 𝐭𝐡𝐞𝐦 𝐦𝐞𝐚𝐧𝐢𝐧𝐠𝐟𝐮𝐥. 𝐇𝐞𝐫𝐞’𝐬 𝐡𝐨𝐰 𝐲𝐨𝐮 𝐜𝐨𝐧𝐯𝐞𝐫𝐭 𝐥𝐨𝐠𝐬 → 𝐬𝐞𝐬𝐬𝐢𝐨𝐧𝐬 → 𝐰𝐨𝐫𝐤𝐢𝐧𝐠 𝐡𝐨𝐮𝐫𝐬

◾ Problem:
Employees may log in and out multiple times a day.
We must pair each login with its next logout, and then compute total working hours per day.

SQL Logic Explained (In Simple Words)

1️⃣ — Pair each login with its next logout
Using LEAD() window function, we fetch the upcoming event_time

2️⃣ — Filter only login rows
Because the logout time is only meaningful for login events.

3️⃣ — Calculate working hours
We subtract login_time from logout_time:
Converts seconds → hours
Handles multiple login-logout pairs in same day

4️⃣ — Summing total working hours per employee per day
This gives us a daily rolled-up summary.


This query is crucial for:
Payroll systems — exact payable hours
Timesheet automation — actual vs expected productivity
Workforce analytics — time-on-platform, shift performance


Day 70 of SQL: https://lnkd.in/gFQ6UgyB


#SQL #Dataanalyst #productanalyst #businessanalyst #sqlqueries

📕Day 72 of SQL Revision
New vs Repeat Customers — A Classic SQL Pattern Every Analyst Should Know

Understanding customer behavior over time is one of most common (& most important) problems in analytics.

One simple but powerful question businesses often ask is 👇
“On each day, how many customers are new and how many are repeat?”

SQL query:

Step 1: Find each customer’s first-ever order
Step 2: Get one order per customer per day --We care about daily active customers, not order volume.
Step 3: Classify customers as New or Repeat
Step 4: Daily aggregation (final output)



WITH first_order AS (
  SELECT
    cust_id,
    MIN(DATE(order_dt)) AS first_order_date
  FROM Input
  GROUP BY cust_id
),
daily_orders AS (
  SELECT DISTINCT
    cust_id,
    DATE(order_dt) AS order_date
  FROM Input
),
customer_status AS (
  SELECT
    d.cust_id,
    d.order_date,
    CASE
      WHEN d.order_date = f.first_order_date THEN 'New'
      ELSE 'Repeat'
    END AS status
  FROM daily_orders d
  JOIN first_order f
    ON d.cust_id = f.cust_id
)
SELECT
  order_date,
  COUNT(CASE WHEN status = 'New' THEN 1 END) AS new_customers,
  COUNT(CASE WHEN status = 'Repeat' THEN 1 END) AS repeat_customers
FROM customer_status
GROUP BY order_date
ORDER BY order_date;



Day 71 of SQL: https://lnkd.in/gj8SwrHk



#SQL #Database #SQLQueries #Dataanalyst #SQLQuery

𝐃𝐮𝐫𝐢𝐧𝐠 𝐣𝐨𝐛 𝐬𝐞𝐚𝐫𝐜𝐡, 𝐭𝐡𝐞𝐫𝐞 𝐚𝐫𝐞 𝐩𝐡𝐚𝐬𝐞𝐬 𝐰𝐡𝐞𝐧 𝐲𝐨𝐮 𝐚𝐩𝐩𝐥𝐲 𝐝𝐚𝐢𝐥𝐲, 𝐩𝐫𝐚𝐜𝐭𝐢𝐜𝐞 𝐝𝐚𝐢𝐥𝐲, 𝐮𝐩𝐬𝐤𝐢𝐥𝐥 𝐝𝐚𝐢𝐥𝐲… 𝐛𝐮𝐭 𝐬𝐭𝐢𝐥𝐥 𝐧𝐨 𝐜𝐚𝐥𝐥𝐛𝐚𝐜𝐤𝐬, 𝐧𝐨 𝐢𝐧𝐭𝐞𝐫𝐯𝐢𝐞𝐰𝐬, 𝐧𝐨 𝐫𝐞𝐬𝐮𝐥𝐭𝐬.

It’s frustrating.
It makes you doubt yourself.
It makes you feel incompatible or not good enough.

But here’s the truth most people won’t tell you:

◾ This phase is normal.
- Almost everyone who made it through faced this exact silence.

Job search is not a straight line.
 It’s effort → waiting → rejection → learning → repeat.

Sometimes it’s not about talent.
Sometimes it’s timing.
Sometimes it’s just probability.

What is in your control?
1) Showing up every day
2) Improving one skill at a time
3) Writing better queries, cleaner logic, sharper explanations
4)  Staying consistent even when motivation drops

Hard work doesn’t always give instant results — but it always compounds.
If you keep learning, keep building, keep trying — the result WILL come.
Maybe not today.
Maybe not this week.
But it will come.


So don’t quit just because the journey feels quiet.
Quiet phases are where growth actually happens.
Give your 100% effort.
Future you will thank you for it.


#Jobsearch #Jobs #Dataanalystsjobs

Day 73 of SQL Revision Series

generate_series() is a set-returning function in PostgreSQL.
This means:
It does not return a single value
It returns multiple rows
Those rows behave like a temporary / virtual table

You can use it in:
SELECT
JOIN
WHERE
GROUP BY
Subqueries

Why is it called a “virtual table”?
Because:
You don’t create a table
You don’t store data
PostgreSQL generates rows on the fly
So it’s fast and very useful for analysis.

Syntax (Very Important):
generate_series(start, stop, step)


| `start`  | First value of the series  |
| `stop`  | Last value (inclusive)   |
| `step`  | Increment / Decrement value |

Analysts Use generate_series():
1. Create Data Without Tables (Practice SQL)
2. Finding Missing Dates in Data
3. Daily / Monthly Reports

generate_series() is used to create a sequence of rows (numbers or dates) without creating a physical table, mainly for analysis and reporting.


Day 72 of SQL: https://lnkd.in/gXZEUhay


#SQL #dataanalysts #sqlqueries #database

Sometimes, It's Not Just Hard Work — Luck Also Plays a Role (And That's Okay!)

We often hear:
“Success = Hard Work + Skills.”
But rarely do people talk about the 3rd variable that silently influences every 
interview:

Luck. Yes, luck matters too.
Let me explain this with a small data-story


📊 Imagine an Interview Like a Random Dataset

You prepare SQL, Python, Product Sense, Case Studies, Stats…
 But the interviewer pulls 3 random questions from a pool of 200+ topics.

Sometimes that sample is perfectly in your favor.
 You get questions that you’ve practiced 100 times.
 You feel confident. You solve everything.
 The interviewer smiles. You smile.
 Result: You get selected.

 Other times…
 The random sample goes against you.
 You get topics you barely touched.
 A trick question comes.
 Your brain freezes for a moment.
 Even though you know you're capable, that day doesn’t reflect it.
Result: Rejection.


But the truth is:
 Your skill didn’t change.
 Only the sample of questions changed.
 Just like randomness in a dataset.



So What Should You Learn From This:
✔ Hard work still matters the most
✔ Don’t judge yourself based on one interview
✔ Rejections don’t mean you’re not good



Day 67: https://lnkd.in/gyADUWw7


#Motivation #CareerMotivation #JobSearch #TechInterview #SuccessMindset #LearningEveryday

Expanding Rows Using CROSS JOIN LATERAL + generate_series()

◾ Why do we need LATERAL: Because the function depends on the column t.n.

LATERAL tells PostgreSQL:
Evaluate this function once per row from left-hand table, & allow it to reference that row’s values."
Without LATERAL, the query either errors out or behaves incorrectly


To repeat each number n times, I used CROSS JOIN LATERAL generate_series(1,n).

LATERAL lets the function access columns from the left table, making this the most efficient and idiomatic PostgreSQL pattern for row explosion.”



“To repeat each number n times, I used CROSS JOIN LATERAL generate_series(1,n). LATERAL lets the function access columns from the left table, making this the most efficient and idiomatic PostgreSQL pattern for row explosion.”


Query:
SELECT t.n AS number_output 
FROM numbers_table t 
CROSS JOIN LATERAL generate_series(1, t.n) AS gs(i) 
ORDER BY t.n, gs.i;


This pattern is used when you want to:
1) Expand one row into multiple rows
2) Generate sequences dynamically
3) Create repetitions
4) Simulate time windows
5) Do simulations (e.g., churn per day, cohort expansion)


Day 69 of SQL: https://lnkd.in/gpA8Efaq



#sql #database #dataanalyst #productanalyst #sqlqueries

This SQL problem looks simple, but it tests how well you understand ordered data, edge cases, and real product logic


Q)Problem Statement:
We have a seat allocation table.
Seats are in order, and we want to swap students sitting in adjacent seats.

Rules:
Seat 1 ↔ 2, 3 ↔ 4, 5 ↔ 6 …
If the last seat is odd, it remains unchanged.
Do NOT change seat numbers — only swap students.

Input:
| seat_id | student |
| 1    | A    |
| 2    | B    |
| 3    | C    |
| 4    | D    |
| 5    | E    |


Query
SELECT
 seat_id,
 CASE
  WHEN seat_id % 2 = 1 THEN
   COALESCE(LEAD(student) OVER (ORDER BY seat_id), student)
  ELSE
   LAG(student) OVER (ORDER BY seat_id)
 END AS student
FROM seats;


Explanation:
Why COALESCE() is Needed
COALESCE(LEAD(student), student)
If next student exists → take it
 Else → keep current student


Day 73 of SQL: https://lnkd.in/g3NyvFtR



#SQL #WindowFunctions #ProductAnalytics #ProductAnalyst #DataAnalytics DataLemur 🐵 (Ace the SQL & Data Interview) Nick Singh 📕🐒 StrataScratch Aditya Kalyan

taplio

The Battle

🥸Hakim Elakhrass

Sachin Kumar

Followers

Posts

Engagements

Likes

Performance Score🥸Hakim Elakhrass's LinkedIn Performance

Performance ScoreSachin Kumar's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

🥸Hakim Elakhrass

Sachin Kumar

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);}🥸Hakim Elakhrass's LinkedIn Performance

Performance ScoreSachin Kumar's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance Score🥸Hakim Elakhrass's LinkedIn Performance