Before and alongside large generative language models, specialized Natural Language Processing (NLP) models—such as fine-tuned BERT/RoBERTa token classifiers for Named Entity Recognition (NER), Sentence-Transformers for dense semantic embeddings, and PCA/UMAP clustering—deliver sub-10ms deterministic inference at 1/100th the cost of generative LLMs.
If you are looking to hire a freelance NLP, Named Entity Recognition (NER), and Transformers programmer, I am Aldian Fazrihady—creator of the Indonesian NER Studio and Visualizing Social Sentiment with Sentence Transformers and PCA.
Production NLP & Token Classification Systems I Build
- Custom Named Entity Recognition (NER): Fine-tuning multilingual and domain-specific biomedical, legal, financial, and Indonesian/English token classifiers.
- Hybrid Dense Embeddings & Clustering: High-throughput Sentence-Transformers pipelines paired with PCA/UMAP topic discovery and `pgvector` semantic search.
- Browser-Native ONNX Runtime Web Inference: Quantizing Hugging Face Transformer models to run directly inside browser Web Workers (Client-Side ML with ONNX & Transformers.js).
Hire a Freelance NLP & Machine Learning Programmer
- Upwork: https://www.upwork.com/freelancers/aldianf
- Email: mobile@aldianfazrihady.com
- Live Demo: Indonesian NER Studio