For employers

Hire this AI Trainer

Sign in or create an account to invite AI Trainers to your job.

Invite to Job
Andre Soeiro

Andre Soeiro

LLM Trainer for Electrical Engineering and Computer Science

Brazil flagBrasília, Brazil
$40.00/hrEntry LevelInternal Proprietary Tooling

Key Skills

Software

Internal/Proprietary Tooling

Top Subject Matter

No subject matter listed

Top Data Types

Computer Code ProgrammingComputer Code Programming
DocumentDocument
ImageImage

Top Task Types

Computer Programming Coding
Evaluation Rating
RLHF

Freelancer Overview

I on-boarded Turing for data labeling on Computer Programming as a Senior Python Developer in September 2024. I was then promoted from an LLM trainer to a pod lead (lead to a team of 6 trainers) then to a calibrator, or team lead, leading 6 teams of 6 trainers each. The project went on pause on January 31st, by the client's request, due to the announcement of DeepSeek.

Entry LevelEnglishSpanishJapanesePortuguese

Labeling Experience

Senior Python Engineer

Internal Proprietary ToolingComputer Code ProgrammingRLHF
* The trainer job involves writing a prompt according to the task's metadata and evaluating the written prompt and the model responses following their criteria on "Instruction Following (If the model does all the prompt asks for)", "Accuracy (if the code has errors or does what is asked for correctly)", "Efficiency (if the code does the task in the best manner or could be implemented better)", "Presentation (If either the code needs better comments or the markdown presentation has issues)", "Up-to-date (if the code uses deprecated libraries or functions)", "Executable Code (If the given code is executable as is by copying it to a file and ran or how much of the code was given by the model)", and "Other issues (any issue in the response that does not belong to the other categories.) * The pod lead does some training workbooks, review workbooks done by trainers. * The calibrator evaluates the reviews from leads.

* The trainer job involves writing a prompt according to the task's metadata and evaluating the written prompt and the model responses following their criteria on "Instruction Following (If the model does all the prompt asks for)", "Accuracy (if the code has errors or does what is asked for correctly)", "Efficiency (if the code does the task in the best manner or could be implemented better)", "Presentation (If either the code needs better comments or the markdown presentation has issues)", "Up-to-date (if the code uses deprecated libraries or functions)", "Executable Code (If the given code is executable as is by copying it to a file and ran or how much of the code was given by the model)", and "Other issues (any issue in the response that does not belong to the other categories.) * The pod lead does some training workbooks, review workbooks done by trainers. * The calibrator evaluates the reviews from leads.

2024

Education

E

Escola Superior de Propaganda e Marketing - ESPM

Professional MBA in Marketing Administration, Marketing Administration

Professional MBA in Marketing Administration
2004 - 2006
U

Universidade de Brasilia

Bachelor's in Electrical Engineering, Electrical Engineer

Bachelor's in Electrical Engineering
1994 - 2002

Work History

C

Centrais Eletricas Brasileiras S.A. - Eletrobras

Planning Engineer

Brasília
2006 - Present