Freelancer logo
Log In Sign Up
Post a Project

EXPLORE

Artificial Intelligence Data Science Deep Learning Machine Learning (ML) Python
Profile cover photo
User Avatar

Kien T.

@tadinhkien

annual-level-two.svg
4.9
252

6.8

6.8

99%

🏆 Python 🏆 AI 🏆 Backend 🏆 Web 🏆 Cloud

$15 USD / Hour

・

Flag of

Vietnam (12:48 AM)

・

Joined on July 7, 2021

・

7 Recommendations

$15 USD / Hour

・

★★★★★ Welcome to My Profile! ★★★★★ I’m Kuro Kien, a Senior Developer with 8 + years of hands-on experience turning ideas into production-ready software. My passion is delivering high-quality, on-time, and on-budget solutions that help you meet your business goals. Below is an updated snapshot of what I bring to the table. ★★ Artificial Intelligence & Data Science ★★★ Generative AI & LLMs - End-to-end chatbot design with OpenAI, Anthropic & open-source models (LLama 2/3, Mistral, Gemma). - Retrieval-Augmented Generation (RAG) pipelines with **vector databases** (Pinecone, Chroma, FAISS, Milvus) for grounded, up-to-date answers. - Multi-agent orchestration (LangGraph / CrewAI) for complex workflows and tool use. ★★★ Computer Vision - Image processing (morphological ops, filtering, segmentation, contour & angle analysis). - Object & lane detection with Mask R-CNN, YOLOv8, Hough Transform, DeepSORT. - OCR pipelines (DB/EAST + Transformer/CTC, Tesseract) and face recognition (MTCNN, FaceNet). ★★★ Traditional Machine Learning - Regression, tree-based models, SVMs, clustering, and feature engineering for tabular data. - Forecasting & Time-Series: ARIMA/SARIMA, Prophet, and deep-learning seq2seq models. ★★★ Frameworks & Tooling: PyTorch, TensorFlow, Keras, scikit-learn, XGBoost, LightGBM. ★★ Data Engineering ★★★ Data Pipelines & Automation - Orchestration with n8n, Airflow & Prefect. - ETL/ELT design, data cleaning, and large-scale web scraping. ★★ Databases - Relational (PostgreSQL, MySQL) & NoSQL (MongoDB, Firestore). - Vector stores for semantic search (Pinecone, Milvus). - Query optimization, indexing, and backup strategies. ★★ Web & API Development ★★★ Backend - FastAPI & Flask for high-performance REST/GraphQL services. - Secure authentication (JWT/OAuth2) and rate-limited public APIs. ★★ Frontend - Interactive dashboards with Dash & Plotly. - React basics for embedding AI widgets. ★★ Full-Stack Delivery - Containerization with Docker & orchestration via Kubernetes. - CI/CD (GitHub Actions, GitLab CI) and cloud deployment (AWS, GCP, Azure, Vercel). ★★ DevOps & MLOps - Model packaging, versioning, and rollout with MLflow & BentoML. - Monitoring (Prometheus, Grafana) and automated retraining triggers. ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------- ★★★★★ How I Work Full-time freelancer** available across time zones (GMT +7). Agile, communicative, and committed to real-time collaboration. Focused on building long-term relationships and measurable success. Let’s team up and turn your next idea into reality—fast, secure, and scalable. Best Regards, Kuro Kien

Verifications

On time

93%

On budget

88%

Accept rate

81%

Repeat hire rate

9%

Freelancer Python Developers Vietnam
See all Python Developers in Vietnam

Portfolio

Portfolio

13132853
I designed and trained deep convolutional neural networks, YOLO to perform precise object segmentation, achieving pixel-level delineation even in highly cluttered scenes. I implemented robust multi-object tracking algorithms that preserve individual identities across frames and handle occlusions and complex motion. Finally, I integrated a high-performance object detection pipeline that rapidly identifies and localizes targets, enabling end-to-end real-time analysis.

Object Segmentation, Tracking, ....

I designed and trained deep convolutional neural networks, YOLO to perform precise object segmentation, achieving pixel-level delineation even in highly cluttered scenes. I implemented robust multi-object tracking algorithms that preserve individual identities across frames and handle occlusions and complex motion. Finally, I integrated a high-performance object detection pipeline that rapidly identifies and localizes targets, enabling end-to-end real-time analysis.

Object Segmentation, Tracking, ....

I designed and trained deep convolutional neural networks, YOLO to perform precise object segmentation, achieving pixel-level delineation even in highly cluttered scenes. I implemented robust multi-object tracking algorithms that preserve individual identities across frames and handle occlusions and complex motion. Finally, I integrated a high-performance object detection pipeline that rapidly identifies and localizes targets, enabling end-to-end real-time analysis.

Object Segmentation, Tracking, ....

13132838
I collected and cleaned the historical time series data, handling missing values, outliers, and seasonal patterns to ensure a robust foundation. I then implemented and tuned multiple forecasting models including ARIMA, Prophet, and an LSTM network and compared their performance using cross-validation metrics. Finally, I selected the most accurate model to generate future predictions and visualized the forecasts to inform strategic decision-making.

Time series forecasting

I collected and cleaned the historical time series data, handling missing values, outliers, and seasonal patterns to ensure a robust foundation. I then implemented and tuned multiple forecasting models including ARIMA, Prophet, and an LSTM network and compared their performance using cross-validation metrics. Finally, I selected the most accurate model to generate future predictions and visualized the forecasts to inform strategic decision-making.

Time series forecasting

13132827
I designed and implemented a scalable retrieval-augmented generation (RAG) pipeline that ingested, indexed, and vectorized diverse document corpora, leveraging FAISS-based similarity search and optimized embedding strategies to ground generated responses in factual external knowledge. I extended the RAG system with agentic capabilities by building a multi-step reasoning agent that dynamically orchestrates tool calls, database queries, and context-driven action plans based on retrieval outputs, ensuring robust execution and error handling. I developed a modular intelligent agent framework combining LLM-based decision-making with integrated API connectors, state management, and safety constraints, iteratively evaluating performance metrics to optimize task completion rates, latency, and user satisfaction.

RAG | Agentic RAG

RAG | Agentic RAG

RAG | Agentic RAG

I designed and implemented a scalable retrieval-augmented generation (RAG) pipeline that ingested, indexed, and vectorized diverse document corpora, leveraging FAISS-based similarity search and optimized embedding strategies to ground generated responses in factual external knowledge. I extended the RAG system with agentic capabilities by building a multi-step reasoning agent that dynamically orchestrates tool calls, database queries, and context-driven action plans based on retrieval outputs, ensuring robust execution and error handling. I developed a modular intelligent agent framework combining LLM-based decision-making with integrated API connectors, state management, and safety constraints, iteratively evaluating performance metrics to optimize task completion rates, latency, and user satisfaction.

RAG | Agentic RAG

13132821
I designed and implemented a scalable data pipeline to collect, preprocess, and manage diverse conversational datasets for training the LLM-based chatbot. I cleaned and standardized the dialogue corpora by handling missing or irrelevant utterances, resolving annotation inconsistencies, and tokenizing inputs to ensure high-quality model data. I conducted exploratory analyses and interactive evaluations visualizing conversational patterns and performance metrics to drive iterative improvements in response accuracy and user engagement.

LLM Chatbot

I designed and implemented a scalable data pipeline to collect, preprocess, and manage diverse conversational datasets for training the LLM-based chatbot. I cleaned and standardized the dialogue corpora by handling missing or irrelevant utterances, resolving annotation inconsistencies, and tokenizing inputs to ensure high-quality model data. I conducted exploratory analyses and interactive evaluations visualizing conversational patterns and performance metrics to drive iterative improvements in response accuracy and user engagement.

LLM Chatbot

13132817
I designed and implemented a scalable data pipeline to ingest and integrate information from multiple sources. I cleaned and preprocessed the datasets by handling missing values, resolving inconsistencies, and standardizing formats to ensure accuracy. I conducted exploratory analyses and generated visualizations to uncover key trends and actionable insights for stakeholders.

Data analysis

I designed and implemented a scalable data pipeline to ingest and integrate information from multiple sources. I cleaned and preprocessed the datasets by handling missing values, resolving inconsistencies, and standardizing formats to ensure accuracy. I conducted exploratory analyses and generated visualizations to uncover key trends and actionable insights for stakeholders.

Data analysis

13132804
I used Stable Diffusion to generate high-quality concept visuals from tailored text prompts. I then imported those assets into Fooocus, where I refined compositions, adjusted color palettes, and applied interface design principles. Finally, I iterated on each piece tweaking diffusion parameters and layout settings to ensure the final designs aligned perfectly with the project’s creative goals.

Stable diffusion and Fooocus Design

I used Stable Diffusion to generate high-quality concept visuals from tailored text prompts. I then imported those assets into Fooocus, where I refined compositions, adjusted color palettes, and applied interface design principles. Finally, I iterated on each piece tweaking diffusion parameters and layout settings to ensure the final designs aligned perfectly with the project’s creative goals.

Stable diffusion and Fooocus Design

13132853
I designed and trained deep convolutional neural networks, YOLO to perform precise object segmentation, achieving pixel-level delineation even in highly cluttered scenes. I implemented robust multi-object tracking algorithms that preserve individual identities across frames and handle occlusions and complex motion. Finally, I integrated a high-performance object detection pipeline that rapidly identifies and localizes targets, enabling end-to-end real-time analysis.

Object Segmentation, Tracking, ....

I designed and trained deep convolutional neural networks, YOLO to perform precise object segmentation, achieving pixel-level delineation even in highly cluttered scenes. I implemented robust multi-object tracking algorithms that preserve individual identities across frames and handle occlusions and complex motion. Finally, I integrated a high-performance object detection pipeline that rapidly identifies and localizes targets, enabling end-to-end real-time analysis.

Object Segmentation, Tracking, ....

I designed and trained deep convolutional neural networks, YOLO to perform precise object segmentation, achieving pixel-level delineation even in highly cluttered scenes. I implemented robust multi-object tracking algorithms that preserve individual identities across frames and handle occlusions and complex motion. Finally, I integrated a high-performance object detection pipeline that rapidly identifies and localizes targets, enabling end-to-end real-time analysis.

Object Segmentation, Tracking, ....

13132838
I collected and cleaned the historical time series data, handling missing values, outliers, and seasonal patterns to ensure a robust foundation. I then implemented and tuned multiple forecasting models including ARIMA, Prophet, and an LSTM network and compared their performance using cross-validation metrics. Finally, I selected the most accurate model to generate future predictions and visualized the forecasts to inform strategic decision-making.

Time series forecasting

I collected and cleaned the historical time series data, handling missing values, outliers, and seasonal patterns to ensure a robust foundation. I then implemented and tuned multiple forecasting models including ARIMA, Prophet, and an LSTM network and compared their performance using cross-validation metrics. Finally, I selected the most accurate model to generate future predictions and visualized the forecasts to inform strategic decision-making.

Time series forecasting

13132827
I designed and implemented a scalable retrieval-augmented generation (RAG) pipeline that ingested, indexed, and vectorized diverse document corpora, leveraging FAISS-based similarity search and optimized embedding strategies to ground generated responses in factual external knowledge. I extended the RAG system with agentic capabilities by building a multi-step reasoning agent that dynamically orchestrates tool calls, database queries, and context-driven action plans based on retrieval outputs, ensuring robust execution and error handling. I developed a modular intelligent agent framework combining LLM-based decision-making with integrated API connectors, state management, and safety constraints, iteratively evaluating performance metrics to optimize task completion rates, latency, and user satisfaction.

RAG | Agentic RAG

RAG | Agentic RAG

RAG | Agentic RAG

I designed and implemented a scalable retrieval-augmented generation (RAG) pipeline that ingested, indexed, and vectorized diverse document corpora, leveraging FAISS-based similarity search and optimized embedding strategies to ground generated responses in factual external knowledge. I extended the RAG system with agentic capabilities by building a multi-step reasoning agent that dynamically orchestrates tool calls, database queries, and context-driven action plans based on retrieval outputs, ensuring robust execution and error handling. I developed a modular intelligent agent framework combining LLM-based decision-making with integrated API connectors, state management, and safety constraints, iteratively evaluating performance metrics to optimize task completion rates, latency, and user satisfaction.

RAG | Agentic RAG

13132821
I designed and implemented a scalable data pipeline to collect, preprocess, and manage diverse conversational datasets for training the LLM-based chatbot. I cleaned and standardized the dialogue corpora by handling missing or irrelevant utterances, resolving annotation inconsistencies, and tokenizing inputs to ensure high-quality model data. I conducted exploratory analyses and interactive evaluations visualizing conversational patterns and performance metrics to drive iterative improvements in response accuracy and user engagement.

LLM Chatbot

I designed and implemented a scalable data pipeline to collect, preprocess, and manage diverse conversational datasets for training the LLM-based chatbot. I cleaned and standardized the dialogue corpora by handling missing or irrelevant utterances, resolving annotation inconsistencies, and tokenizing inputs to ensure high-quality model data. I conducted exploratory analyses and interactive evaluations visualizing conversational patterns and performance metrics to drive iterative improvements in response accuracy and user engagement.

LLM Chatbot

13132817
I designed and implemented a scalable data pipeline to ingest and integrate information from multiple sources. I cleaned and preprocessed the datasets by handling missing values, resolving inconsistencies, and standardizing formats to ensure accuracy. I conducted exploratory analyses and generated visualizations to uncover key trends and actionable insights for stakeholders.

Data analysis

I designed and implemented a scalable data pipeline to ingest and integrate information from multiple sources. I cleaned and preprocessed the datasets by handling missing values, resolving inconsistencies, and standardizing formats to ensure accuracy. I conducted exploratory analyses and generated visualizations to uncover key trends and actionable insights for stakeholders.

Data analysis

13132804
I used Stable Diffusion to generate high-quality concept visuals from tailored text prompts. I then imported those assets into Fooocus, where I refined compositions, adjusted color palettes, and applied interface design principles. Finally, I iterated on each piece tweaking diffusion parameters and layout settings to ensure the final designs aligned perfectly with the project’s creative goals.

Stable diffusion and Fooocus Design

I used Stable Diffusion to generate high-quality concept visuals from tailored text prompts. I then imported those assets into Fooocus, where I refined compositions, adjusted color palettes, and applied interface design principles. Finally, I iterated on each piece tweaking diffusion parameters and layout settings to ensure the final designs aligned perfectly with the project’s creative goals.

Stable diffusion and Fooocus Design

4.9 · 252 Reviews
252 Reviews
User Avatar
Flag of

Anuj T.

5.0

₹9,000 INR

1 month ago

CLIP vs SigLIP Experiment (archived)

Really easy to work with. Good communication and hardworking.

Share

H
Flag of

Hamad A.

5.0

$130 USD

1 month ago

Artificial intelligence LLM

great work, on time

Artificial Intelligence
Large Language Models (LLMs)

Share

User Avatar
Flag of

Deepak A.

5.0

₹8,925 INR

4 months ago

Part-Time Python Prototype Enhancement

good work and timely feedback

JavaScript
Python
Django
Software Architecture
Software Development

+5 more

Share

T
Flag of

Nikola B.

5.0

$120 USD

8 months ago

Python Hybrid Movie Recommender

Kien is very fast, reliable and has such a good communication. We will work more for sure!

Python
Data Processing
Software Architecture
Machine Learning (ML)
Data Mining

+3 more

Share

User Avatar
Flag of

Razan H.

5.0

$125 USD

9 months ago

Liver tumor segmentation and detection

Working with Kien T was a great experience. He is professional, responsive, and highly skilled. He understood the project requirements clearly and delivered quality work on time. Communication was smooth, and he was always open to feedback and improvements. I would definitely recommend him to others and look forward to working with him again

Python
Machine Learning (ML)
Data Science
Artificial Intelligence

+1 more

Share

Experience

AI Engineer II

Jul, 2019 - Present

•

6 years, 10 months

Microsoft

Jul, 2019 - Present

•

6 years, 10 months

AI Engineer

Flag of

Ho Chi Minh, Vietnam

Jul, 2019 - Present

•

6 years, 10 months

Education

University of Michigan - School of Information

2014 - 2016

•

2 years

Master

Flag of

United States

2014 - 2016

•

2 years

Certifications

P

Preferred Freelancer Program SLA

Freelancer logo US (International) / English
Help & Support

Freelancer

Categories Projects Contests Freelancers Enterprise Innovation Challenges AI Development Membership Preferred Freelancer Program Project Management Local Jobs Photo Anywhere Showcase API for Developers Get Verified

About

About us How it Works Security Investor Sitemap Stories News Team Awards Press Releases Careers

Terms

Privacy Policy Terms and Conditions Copyright Policy Code of Conduct Fees and Charges

Partners

Escrow.com Loadshift Warrior Forum

Apps

Apple App Store logoGoogle Play logo

Registered Users

Total Jobs Posted

Freelancer ® is a registered Trademark of Freelancer Technology Pty Limited (ACN 142 189 759)

Copyright © 2026 Freelancer Technology Pty Limited (ACN 142 189 759)