Amazon Mechanical Turk Api Jobs

4 were found based on your criteria {{ paging.total|number:0 }} were found based on your criteria

show all
  • Hourly ({{ jobTypeController.getFacetCount("0")|number:0}})
  • Fixed Price ({{ jobTypeController.getFacetCount("1")|number:0}})
Hourly - Expert ($$$) - Est. Time: More than 6 months, 30+ hrs/week - Posted
- Will have strong experience with multiple Amazon Web Services IoT, S3, Dynamo DB, Lambda Functions, Mechanical Turk, and SNS - Raspberry Pi experience - Will recommend optimal architecture for solution - Can start immediately - Highly available & reachable on Skype - Any experience with machine learning image classification APIs a big bonus/plus
Skills: Mechanical Turk API AWS Lambda Raspberry Pi
Fixed-Price - Intermediate ($$) - Est. Budget: $500 - Posted
Description Web application that allows Amazon Turks to create content in two phases. Phase 1 allows Turks to create content and receive immediate feedback as to whether the content is unique or similar to phrases we already have in our database. Phase 2 should allow Turks to rate content they have not rated yet and provide a rating of the phrase. The final phrase will aggregate the ratings and identify top content makers by their crowd sourced rating. Using the data created I will manually rank Turks to a qualified type and invite them to higher paying hits. Technical Requirements: All code should be developed in Python and use AWS Boto. The web piece should run in Elastic Beanstalk and allow me to load into EB and stop and start it. NoSQL Database should be used. Application 1: Build a website and process which allows Turks to submit phrases and store them in a nosql database and stores and associates the turk’s id the phrase. The application should: Verify the phrase does not already exist in the database. We will need some sort of logic to determine the phrase is near similar to an existing phrase. The website should show the Turk what phrases it feels are similar to the phrase already submitted. And remind the Turk unique phrases have a better chance to be higher rated in the next phase. The Turk will need to “categorize” their phrase into one to five categories. Categories will be loaded from a JSON file during web server startup. Award the Turk a HIT for the phrase submitted. The administrator of the application should be able to start a “project” in AWS Turk and through configuration tell Application 1 to use a specific AWS Turk project. The application will award a HIT to a Turk for every phrase submitted even if they are duplicated. We might have a third application later for Turks to find near duplicates and sort out Turks that make dupes. Or we might develop search logic in DynamoDB that has a rating as to how similar the phrases are. Application 2: Website which allows turks to rate the phrases in the nosql database. The rating should be stored in the nosql database and store the Turk’s id doing the rating. The id is stored so as the Turk rates more phrases it only pulls back phrases they have not rated. Phrase Rating should be: Luke warm Not bad Really good Really hot Really really hot. OMG I don’t believe this text. Another configuration item should be how many ratings per phrase to allow. This should be a config item pulled from a json file during load of the application. For example we start the application and allow up to 20 Turk’s to rate a phrase. After analysis we determine we do not have enough Turks to definitely determine how accurate the rating is. So we restart this and up it to 100 Turks who can rate a phrase. The rating scale should be dynamic and should be loaded from a json file I can change for each project. This allows me to change the rating scale and later rating process. As Turks Rate the phrases it should store what version of the rating is currently loaded. If I start the application with the above rating example and later want to change it. I want to be able to “sort” out the ratings and what rating scale was used. Once a Turk submits a rating they will be rewarded a HIT. The administrator of the application should be able to start a “project” in AWS Turk and through configuration tell Application 2 to use a specific AWS Turk project. Application 3: What To Do With the Rating: The following python job (ie I can run this code on demand) will promote a Turk into a specified qualification type: All of the following must be true to promote Turk to specified qualification type: 20 individual turks rate one phrase at “really good” or greater. Five or more phrases created by that turk are rated “really good” or greater by 60 or more Turks. Each phrase has to have 5+ Turks rate each phrase as high Example: phrase 1 rated high by 10 people, phrase 2-4 rated high by 5 people, phrase 5 rated high by 20 people. The number of people required for each “rule” should be dynamic. I would like to change a JSON file if I want to change the number of people required for each rule. Ie. Instead of 20. I want to up it to 100.
Skills: Mechanical Turk API Python
Fixed-Price - Intermediate ($$) - Est. Budget: $300 - Posted
I have a database that contains information about a number of properties that will be going to auction. The database that I used is called Airtable and they have an API available. I need mechanical turk to pull data from my database to create new hits for turkers to complete. Their job will be to make a phone call and update the record. This updated data needs to be then pulled back into the database. I need the mechanical turk hits to be created based on specific criteria or rules automatically. Please explain in as much detail as possible how you plan to complete the task. Please be responsive because I'm looking to hire somebody in the next 24-48 hours.
Skills: Mechanical Turk API API Development
Hourly - Entry Level ($) - Est. Time: Less than 1 month, 10-30 hrs/week - Posted
Looking for a backend developer to: 1. Take .txt email files available here: https://www.cs.cmu.edu/~./enron/enron_mail_20150507.tgz, 2. store them in redshift database then 3. parse each .txt file to identify 'from', 'to' and 'subject' and create three columns in the database. After the data is stored in redshift: 1. Connect to Mechanical Turk API 2. Automatically create an Mturk HIT to have workers review email subjects and categorize words into categories (i.e. verb, action item, due date) 3. Store HIT results in a new column in the redshift database.
Skills: Mechanical Turk API Amazon S3 HTML5 SQL