The best works of freelancers in machine learning on Freelancehunt — examples of work performed in machine learning ― page 2

Switch to English version?

Yes

Переключитись на українську версію?

Так

Переключиться на русскую версию?

Да

Przełączyć się na polską wersję?

Tak

Login
Registration
- Welcome to Freelancehunt
  
  Work risk-free, saving time and money
  
  Login Registration

21 000 USD
Granite Analysis & Segmentation & Classification

AI & Machine Learning

Comprehensive development of the automation system of marble production.
Development of the detection model on the tape and the classification of the marble plate.
Automatic calibration of the camera and the measurement of the parameters of the plate (type classification, size measurement), contour assessment and anomalies.
Segmentation of the image, setting virtual backgrounds.
… The system is operated on 4 manufacturing sites in Italy and Austria.
The general type:
HTTPS://youtu.be/juHldd5llHg

Technical points:
HTTP://pot.pp.ua/video/granitev3.m4v

webitalex
2500 USD
FaceID: Face Detection and Identification

AI & Machine Learning

A Python and C++ library for the detection and recognition of persons.

Continuous before learning model in the process of recognition.

… The video is available to the top of the portfolio in the resume section!
https://www.youtube.com/watch?v=P45HTL08IaM

Face detection 0.04 ~ 0.1 sec/frame
Learning features of each face ~0.2 sec/face on avr CPU
Face Recognition:
High accuracy more then 97.5%, Sensitivity ~ 75%
Very fast searching in faces database (500k faces per second)
Live learning during the recognition process

webitalex
20 USD
Creation of a personal character with artificial intelligence

AI & Machine Learning

Goal: A unique girl with artificial intelligence from two ethnic groups (without a real person, without LoRA).

Solution: Combined two photographs using Seedream AI. Adjusted prompts to merge facial features. Selected the best result + retouching in Photoshop.

… My role: Prompts, artistic direction, post-processing.

Skills: Artificial intelligence generation (Seedream, Flux), prompt engineering, Photoshop.

Result: A unique hyper-realistic female character. Ready for brand/commercial use.

#model #AI-content #ai #human #realism

anastasiia_nesprava
1200 USD
AI Assistant for Call Automation in a Dental Clinic

AI & Machine Learning

An automation system was implemented to process incoming and outgoing calls for a dental clinic. The solution not only records every call but also converts it into structured data for further work with patients and quality control of administrators. The system receives call data from Binotel, performs speech-to-text transcription, analyzes conversations, sends results to Cliniccards CRM, and generates weekly analytics for all interactions.

How the solution works:

… 1) The system automatically receives data for every incoming and outgoing call from Binotel.
2) Call recordings are automatically converted into text for further analysis.
3) Based on the conversation, the system identifies the main intent:
appointment booking, consultation, rescheduling, interest in specific services, and other details.
4) Data is automatically transferred to Cliniccards CRM:
- If a patient does not exist — a new record is created.
- If the patient already exists — their data is updated.
5) The system evaluates call quality based on predefined criteria:
greeting, clinic introduction, правильні питання, correct closing of the conversation.
6) Weekly reports are generated with key metrics:
number of new patients, most common requests, communication quality, and other KPIs.

Key business benefits:

- Improved call handling quality
No calls are missed or left unanalyzed. Management gains full visibility of staff performance.

- Automatic CRM data population
All call data is automatically saved in patient profiles without manual input.

- Staff performance control
The system evaluates communication quality and helps identify improvement areas.

- Deep call analytics
The clinic receives insights into patient needs, popular services, and new client flow.

- Time savings
No need to manually listen to calls or update CRM — everything is automated.

Conclusion:
Call automation enabled the clinic to combine telephony, AI analysis, and CRM into a single system. The business now has full control over communication, structured data, and actionable analytics for decision-making.

#AI #AIAgent #AIAssistant #OpenAI #GPT #n8n #Makecom #Automation #BusinessAutomation #Binotel #Ringostat #Twillio #CRM #CallAnalysis #SpeechToText #Transcription #DentalClinic #Automation #calltracking

d4rk51d3n
135 USD
Synchronization architecture: Notion → Reclaim.ai

AI & Machine Learning

Goal: Automate the scheduling of production tasks from Notion to Reclaim.ai via Google Tasks. Key challenge: implement reliable deduplication without changing statuses in Notion and bypass strict Google API quotas.

Solution: Two-level architecture on n8n (Railway):

… 1. Collection and Validation (Notion Pipeline)
Smart Deduplication: The system ignores duplicate Notion triggers by matching page IDs with its own database (Data Table). This allows the status PRODUCE to remain static.

Dynamic Time-Window: Filtering tasks by a 14-day window (REZERWACJA), which excludes scheduling of archived records.

JS Hours Parser: Code Node automatically converts free input of hours into the format (duration: Xh), understandable for Reclaim.ai AI algorithms.

2. Delivery and Optimization (Queue Engine)
Quota Management: Implemented batching and retries, which eliminated 403 Quota Exceeded errors during bulk operations with Google API.

Asynchronous Flow: Distribution into "Collector" and "Sender" through a status queue (PENDING -> SENT), ensuring 100% delivery of each task.

Result:

Sync Speed: Appearance of the task in the calendar within 1–5 minutes.

Stability: Complete automation without "manual" support of statuses in Notion.

Scalability: Ready infrastructure for scaling to other departments of the company.

#n8n #Notion #ReclaimAI #Automation #Backend #API

Glovinskiy
7 USD
Quick creation of Shi content

AI & Machine Learning

Quickly create a video on your topic)

anastasia_khuhaieva
564 USD
Auto-posting to 16+ social networks based on n8n

AI & Machine Learning

Task
It was necessary to automate the regular auto-posting of content on social media to avoid manual uploading of posts every day. It was important to pull content from a table/drive, check that the post had not yet been published, adapt it for different platforms, and have a transparent history of executions.

Solution
… A multi-step workflow was assembled in n8n, which is triggered on a schedule (Schedule Trigger) and goes through the full cycle from selecting posts to their publication. At the start, the script pulls a list of scheduled posts from a table/Google Sheets, filters records by publication date and status, and checks for the presence of media files in cloud storage. Next, branching is set up: separate branches are created for each post for different platforms (for example, LinkedIn, Facebook, Instagram), where text formatting, addition of UTM tags, and image uploads take place. At the final nodes, publication is decided: n8n sends the post to the corresponding social media API, changes the post status in the table to "published," and logs the execution result (success/error) for further analysis.

Result
The content plan is now executed completely automatically — it is enough to add a new entry to the table, and the post enters the auto-posting queue according to the specified schedule. The team has stopped wasting time on manual publication, and the risk of "forgetting to post" specific material has been reduced to almost zero. All posts now have a uniform structure, correct links, and UTM tags, and through the log in n8n, it is easy to track what exactly was published and where errors might have occurred.

In numbers
- 1 universal workflow in n8n covers auto-posting to several platforms at once.
- Up to 90% of routine operations for posting have been automated.
- Savings of up to 10 hours per week on manual uploading and formatting of content.
- 100% of scheduled posts are published according to the content plan, with no omissions due to human factors.

Oleh_marechko
5 USD
Visualization of the furniture that we manufacture

AI & Machine Learning

Visualization of a soft chair for children

Taras_Radchuk
Knight

AI & Machine Learning

Advertising animation

Kritt
350 USD
Intelligent Control System Development for Industrial Batteries

AI & Machine Learning

Goal: Maximizing profit from electricity arbitrage on the "Day-Ahead Market" (DAM).

Business Challenge

… The client operated a SmartLogger 3000C01 industrial battery with a 400 kWh capacity but managed it manually. The objective was to create an automated system capable of:

Analyzing hourly electricity prices on the DAM (Day-Ahead Market).

Accounting for the facility's real-time power consumption.

Creating an optimal charge/discharge schedule.

Maximizing revenue from electricity sales.

Technical Implementation

Tech Stack:

Core: Python 3.x (Flask, SQLite)

AI: Google Gemini AI API (gemini-2.0-flash-exp)

Integrations: SOAP API (SmartLogger), REST API (OREE - Energy Market), Excel Parsing

Automation: Cron

System Architecture:

1. Data Collection Module:

Integration with OREE API to fetch DAM prices for the next day.

Parsing historical consumption data from Excel (KWT.xls).

Reading current battery status via SmartLogger SOAP API.

2. AI Optimizer (System Core):

Development of a specialized prompt for Gemini with a step-by-step algorithm.

Analysis of a 24-hour window considering:

Hourly prices (UAH/kWh).

Forecasted facility consumption.

Technical constraints (charge/discharge rates).

ROI threshold (minimum margin of 3 UAH/kWh).

Support for multi-cycle optimization (morning + evening peaks).

Adaptive discharge based on actual consumption.

3. Execution Module:

Automatic schedule execution via SOAP API.

Hourly monitoring and adjustment.

Operation logging and Telegram status notifications.

4. Web Interface (Flask):

Dashboard with performance visualization.

Operation history, profit stats, system configuration, and access control.

Results

Technical Achievements:

Increased Discharge Hours: From 2 to 10 hours per day.

Profit Growth: 11% increase (from 2,874 to 3,198 UAH/day).

Automation: 100% of routine operations automated.

Forecast Accuracy: 95%+.

Economic Impact:

Projected Monthly Profit: ~96,000 UAH.

System ROI: Payback period of 2–3 months.

Time Savings: 2–3 hours saved for the client daily.

Key Technical Solutions

AI Integration: Custom prompt engineering, JSON Mode for guaranteed response structure, and fallback mechanisms for AI unavailability.

Consumption Optimization: Analysis of historical data from the previous week and consideration of the facility's daily work schedule.

Reliability: Retry mechanisms for API requests (up to 10 attempts) and backup scenarios for connection failures.

Automation: Cron jobs for daily forecasting (00:00) and 24/7 continuous operation.

Implementation Complexity: SOAP/REST integrations, dynamic programming algorithms, production deployment (SSH, Linux). Uniqueness: Hybrid approach (AI + Business Logic), adaptability to real-time consumption rather than theoretical maximums, production-ready autonomy.

Skills Applied: Python, AI/ML Integration, Google Gemini API, SOAP/REST API, Flask, SQLite, Cron, Automation, Excel Parsing, Production Deployment, Linux Administration, Algorithm Optimization, Data Analysis, Industrial IoT.

Duration: 2 weeks | Role: Full-stack Developer + AI Integration | Status: Live in production, autonomous

maciborka
Comprehensive architecture and automation GoHighLevel

AI & Machine Learning

Task: Centralize the management of potential clients and automate operational workflows for a group of dental laboratories to prevent the loss of potential clients.

Solution: Created a GoHighLevel ecosystem with 2 accounts and dedicated email domains (SPF/DKIM). Integrated FB Lead Ads, telephony, and chat. Developed custom pipelines and "Speed-to-Lead" automation to recover missed calls and no-shows.

… Result: A ready CRM system that optimized case intake, ensured 100% capture of potential clients, and improved response time through automation.

ivan_kyslov
AI specialist portfolio

AI & Machine Learning

Portfolio where you can view my works

Yuliia_Sofiichuk

AI & Machine Learning