Sign In

Daoud Clarke

Machine Learning Engineer at Lucid, LLC

Daoud Clarke is a Machine Learning Engineer at Lucid, LLC, where he has been employed since 2020. He possesses a strong background in software engineering and machine learning, demonstrating the ability to analyze problems swiftly and devise effective solutions.23

In addition to his role at Lucid, Clarke co-founded DataPastry, a data science venture, and worked as a data scientist on a contract basis for Tesco from 2018 to 2019.3 He is also involved in the academic community, having presented at conferences such as the 6th International Open Search Symposium in October 2024, where he discussed topics related to machine learning and search engine development.1

Clarke's professional profile highlights his expertise in various areas of machine learning, analytics, and software development, making him a notable figure in his field.2

Highlights

Dec 3 · linkedin

Have you ever seen big data tools being used when they shouldn't be? I have. Again, and again, and again. And it's not pretty.

Once data scientists learn Spark they think they have to use it for everything. You don't. And just because your data is in Hive/$BigDataStore, you don't have to keep it there. In fact, I like to get it out as quickly as possible!

Here's some tips for handling big data (full article: https://lnkd.in/dFdWtWq ):

  1. Take some time to think about the type of problem you're solving before you start

  2. Figure out how much data you actually need

  3. Is your dataset highly biased?

  4. Think about the best way to access the data for the problem you're solving

  5. Where is the best place to store the data? Can it be in memory or can you copy it to a local disk?

  6. If it doesn't fit in memory, consider getting a VM with more memory instead of spending time engineering a solution.

  7. Don't use Spark - unless you have to! Most of the time, you don't.

  8. Consider using Dask when you do need to parallelise.

Anything else? Let me know in the comments!

#datascience #bigdataanalytics #hadoop

Nov 27 · linkedin

"Most data scientists don’t generate quantifiable business value."

A controversial statement from my cofounder Iskander Yusof. Here's what else he has to say:

Occasionally you’ll hear claims like “I was responsible for a 10% increase in user engagement” but when you look more closely you’ll quickly find that these statements just don’t hold up.

In our latest article, we explain how to get the most value from a data scientist: https://lnkd.in/eZzQCMt

If you’ve just hired a data scientist, you should have a plan for onboarding them to get them up to speed as quickly as possible. Make sure your data scientist understands the problem(s) you want to solve and the wider business context, ensure they have easy access to the right people and -- obviously -- give them access to the data they need. Most of these things should be common sense but are easier said than done!

It’s also critical to “facilitate, not dictate” when working with a data scientist. We’ve put together a little Venn diagram to give an idea of how responsibilities might be divided.

How do you ensure data scientists deliver value in your own organisation? Are you able to quantify what they bring to the table?

#datapastry #datascience #analytics

Related Questions

What projects has Daoud Clarke worked on at Lucid, LLC?
How did Daoud Clarke transition from DataPastry to Lucid, LLC?
What are Daoud Clarke's main areas of expertise in machine learning?
Can you provide examples of Daoud Clarke's contributions to Meta-Press.es?
What is the MWMBL.org project that Daoud Clarke is involved with?
Daoud Clarke
Daoud Clarke, photo 1
Daoud Clarke, photo 2
Add to my network

Location

Swansea, Wales, United Kingdom