OpenAI Model Turning
113 USD We are developing a platform for assessing the fire resistance of homes in the USA. There are implemented algorithms that allow for evaluation and generate a list of recommendations to improve such fire resistance.
At this stage, we need to implement auxiliary mechanisms based on OpenAI models to help users select the most important recommendations from the list. The models are already configured based on ChatGPT, now we need to transfer them to the business account for API access, and also develop/configure embeddings to reduce token consumption.
We do not have a specialist on our team with sufficient experience working with OpenAI & Embeddings, so we are looking for a specialist for collaboration in this area.
Please, when submitting your application, indicate what commercial experience you have in implementing similar tasks.
-
7 days113 USD
1654 11 0 7 days113 USD
Good day! This is a very interesting task. I see that you are looking for not just an API integrator, but a specialist who will help you build an effective and cost-efficient architecture for working with LLM for prioritizing recommendations. Regarding your request about embeddings to reduce token consumption — this is absolutely the right direction. This approach, known as Retrieval-Augmented Generation (RAG), allows for significant cost optimization by not sending the entire array of recommendations in each request, but only the most relevant sample.
I propose the following implementation plan:
Step 1. Audit. Analyze your existing algorithms and prompts currently used in ChatGPT.
Step 2. Create a vector database. We will take your entire list of recommendations, create an embedding for each of them, and upload them to a vector database (for example, Pinecone or Supabase).
Step 3. Build search logic. Implement a mechanism that converts the user's home input data into a vector and performs a semantic search in the database to obtain relevant recommendations.
…
Step 4. Integration with LLM. Create the final prompt that passes the obtained recommendations to the main model for final prioritization and response generation for the user.
Step 5. Transfer to the business account. Set up access via API from your OpenAI business account and conduct final testing.
For a more accurate estimate, I would like to clarify a few points:
- What is the approximate volume of your recommendation database (number and average text size)?
- Where and in what format are these recommendations currently stored?
- Do you have any preferences regarding the hosting of the vector database?
I am ready to discuss the details in correspondence.
-
3 days113 USD
412 1 0 3 days113 USD
Good day! I have experience in integrating and setting up work with OpenAI. Please write to me in private messages to discuss your task in more detail. Thank you!
-
14 days113 USD
223 2 0 14 days113 USD
Good day. I have a lot of experience with LLM models, including OpenAI. Could we discuss the task in more detail?
Current freelance projects in the category AI & Machine Learning
Improve the performance of Claude Code and work on software development.I am currently developing CRM and Analytics software. I am using Claude Code, but I understand that the results are not the best in terms of changes. There are 2 tasks - Need help creating a preset for skills, MD, and so on to improve quality. Take verified ones that have been… AI & Machine Learning, Python ∙ 25 minutes back ∙ 5 proposals |
I am looking for a developer of AI agents and automation for business.
1000 USD
We are looking for agents for: — Sales and lead qualification — Marketing and content — Customer support — Analytics and reporting — Internal processes (HR, knowledge base) — Specialized niches (real estate, medicine, legal services) If you have experience or ready solutions —… AI & Machine Learning, Bot Development ∙ 43 minutes back ∙ 11 proposals |
A specialist is needed for training a model for object detection in architectural drawings.1. Task ObjectiveIt is necessary to train or adapt a model for the automatic detection and classification of elements in architectural drawings of interior projects.The system should find objects on the plans, outline them with a bbox/contour, determine the class of the object,… AI & Machine Learning ∙ 1 hour 16 minutes back ∙ 2 proposals |
AI agent in Telegram for online store
451 USD
Task: creating an AI agent in Telegram for live communication with customers and processing orders in the online store on OpenCart. It is also necessary to implement customer notifications regarding order statuses, store operating hours, working conditions, delivery times, and… AI & Machine Learning, Bot Development ∙ 14 hours 19 minutes back ∙ 51 proposals |
Improvement of the website on Claude Code
180 USD
WE are creating a website for real estate publication. The main part of the project, the landing page, and the admin panel are already in place. Everything is functioning correctly, but there are improvements that require attention and professionalism. We will provide more… AI Art, AI & Machine Learning ∙ 19 hours 46 minutes back ∙ 40 proposals |
Vinnytsia