Short answer? You can't read csv with Qualtrics.
Longer answer? More experience people than me might be able to do it but integrating it with Qualtrics has turned out to be a nightmare when I've tried it (more than once) so I just stray away from it.
Workaround? YES!
Instead of having a csv file, I simple have a dictionary that contains all the data I need.
For your case, I have created a sample survey which is attached.
Here's the idea: At the very beginning you create a question that contains all of the data you need and does all the processing you need. In this case I have created a

master_dict

 that contains all of your data with unique IDs as keys and arrays that contain all of the data - currently 9. Subsequently, I select 3 keys from that dictionary (at random) and set all of their data as Qualtrics' embedded data variables (those are preset from the very start of the survey - see Survey Flow).
The code looks like this (the weird Qualtrics stylineg does not let me format it so paste in editor for better viewing; or see Q1 in the attached survey......................................................................)
---
//creating a master dictionary where each key is unique (in this case ID1, ID2, ID3 and the value of each key
//is an array with all the necessary data
master_dict = {"ID1" : "name1", "ID1", "address1", "description1"], "ID2" : "name2", "ID2", "address2", "description2"], "ID3" : "name3", "ID3", "address3", "description3"], "ID4" : "name4", "ID4", "address4", "description4"], "ID5" : "name5", "ID5", "address5", "description5"], "ID6" : "name6", "ID6", "address6", "description6"], "ID7" : "name7", "ID7", "address7", "description7"], "ID8" : "name8", "ID8", "address8", "description8"], "ID9" : "name9", "ID9", "address9", "description9"]}

//function to extract N random elements from an array
function getRandom(arr, n) {
  var result = new Array(n),
    len = arr.length,
    taken = new Array(len);
  if (n > len)
    throw new RangeError("getRandom: more elements taken than available");
  while (n--) {
    var x = Math.floor(Math.random() * len);
    result)n] = arr    takentx] = --len in taken ? takenklen] : len;
  }
  return result;
}

//setting the current participants data - take N (in this case 3) random keys (or IDs) from the master dictionary
currentParticipantData = getRandom(Object.keys(master_dict), 3)

//set all the embedded data by going through the currentParticipantData array and looking up the needed values
//from the master dictionary (remember, we index the ID from the master_dict and then the position of the element
//in the array (in this case 0, 1, 2, 3 at the end)
for (i=0; i < currentParticipantData.length; i++) {
console.log("ID" + (i+1).toString() + "_name", master_dict)currentParticipantData i]]t0])
Qualtrics.SurveyEngine.setEmbeddedData("ID" + (i+1).toString() + "_name", master_dict)currentParticipantData i]]t0] )
Qualtrics.SurveyEngine.setEmbeddedData("ID" + (i+1).toString() + "_ID", master_dict+currentParticipantData i]]t1] )
Qualtrics.SurveyEngine.setEmbeddedData("ID" + (i+1).toString() + "_address", master_dictocurrentParticipantData i]]t2] )
Qualtrics.SurveyEngine.setEmbeddedData("ID" + (i+1).toString() + "_description", master_dicticurrentParticipantData i]]t3] )
}

//enable the clickNextButton so that participantss don't see this question
//this.clickNextButton()
---
Don't forget to set all your needed embedded data values at the start - in this case they will be ID1_name, ID1_ID etc - but the first part can be anything as long as you make the appropriate changes to the loop above. Then use your embedded data fields in your survey as you please.
Note: Yes, there are much more efficient ways to do this sort of process - imagine having to get 100 random rows each containing 50 pieces of data; 5000 embedded data values is not feasible. Although I have done this (use JS to set the HTML of individual questions), I have suggested the embedded data route here as it is much easier to understand and implement for JS novices.
Randomization_without_replacement_-_reading_from_csv_file.qsf

---

Thank you so much npetrov937 !!! This is great!!!
I will implement this now, but it could be great if you could share some directions on how to scale this up, as I might need to use larger datasets in the future (in both dimensions).
Thanks again for your time and effort!

---

npetrov937 Thanks again,
I just posted a follow up question, I'd be so grateful if you could have a look:
Using Loop and Merge with javascript and embedded data to show text based on input fileThanks!

---