Tien N

Multi-Tenant Support Chatbot Platform

Nexcel Solutions • Temps plein

Nov 2025 - Present • 6 mos

Key Tasks: 1. RSR-ReportSynthesizer: Enhance and refactor the entire project following SOLID principles to support chatbot capabilities, including automatic chart generation for data visualization, parallel chart generation using multi-threading and locks, conversational memory, interactive querying of DataFrame data, chart-based insights, chart editing, and user assistance with data analysis. 2. Generative AI Tools: - Multi-agent architecture: Each agent is specialized for a distinct responsibility (context analysis, planning, execution, insights, etc.). - Provider redundancy: Supports Groq, Gemini, OpenAI, Cerebras, OpenRouter, Cohere, and Ollama to avoid vendor lock-in. - Model fallback & priority: Models are tried in priority order using a round-robin strategy. Task-aware model selection: Lightweight models are used for fast chatbot responses, while larger reasoning models are reserved for complex analysis. - Fault tolerance: API key rotation, retry control, and provider-level timeouts ensure stability. 3. Full-stack Implementation: - Frontend: Proficient in HTML, CSS, and JavaScript to build responsive user interfaces. - Backend: Developed AI-driven APIs in Python following SOLID design principles to ensure scalability, maintainability, and extensibility. - Database: MySQL, Redis.

Full-Stack AI Engineer

GoBuy Vietnam • Temps plein

Oct 2024 - Oct 2025 • 1 yr

1. CV Screening & Matching – Built an NLP pipeline with spaCy + Hugging Face Transformers to extract candidate data, applied rule-based + embedding similarity scoring, and deployed as a Django REST API with PostgreSQL. 2. Facial Emotion Recognition – Trained CNN models in PyTorch, integrated with OpenCV for real-time video streams, and optimized inference with ONNX Runtime. 3. AI Chatbots – Fine-tuned LLaMA, Gemma, ChatGPT-4o-mini; implemented context memory with Redis + PostgreSQL; deployed APIs via FastAPI for multilingual customer support. 4. Meeting Recording & Summarization – Automated transcription with Whisper-1, generated summaries using BART/Pegasus, and deployed as microservices with Celery + RabbitMQ. 5. Generative AI Tools – Delivered Stable Diffusion / ControlNet / AnimateDiff for text-to-image, image-to-image, and video generation; implemented background removal (U²-Net) and enhancement (ESRGAN) features. 6. Full-stack AI Apps – Developed responsive frontend (HTML, CSS, JS), backend APIs (Python, Django, FastAPI), and PostgreSQL databases for CVs, chatbot logs, and AI media storage. 7. Data Scraping & Analytics – Built TikTok scraping pipelines using Playwright, BeautifulSoup, and AsyncIO; stored metadata in PostgreSQL; created Power BI dashboards for engagement trends and reporting.

Data Scientist

Puritype • Temps plein

Jan 2023 - Sep 2024 • 1 yr 8 mos

At Puritype, I worked remotely as a Data Scientist, focusing on improving recommendation systems. I built MVPs, ran hypothesis tests, and performed A/B testing to find the best approaches for our models. I used AI and machine learning to make the recommendations more accurate, and we regularly gathered client feedback to keep improving the system. I also integrated large language models (LLMs) to create more personalized experiences for users. Throughout the project, I worked closely with the team to make sure everything aligned with our business goals.