NLP Internship – spaCy & Custom NER (Pre-Junior / Production-Ready Only)

Important — Read Before Applying

This is NOT a beginner NLP role.
If you have only used pre-trained models or followed tutorials, do not apply.

You must already know how to train, evaluate, and improve custom NLP models, specifically Named Entity Recognition (NER) using spaCy.

Required Experience (Mandatory)

You will be automatically rejected if you do not have:

• At least 1 real NLP project involving custom NER
• Hands-on experience with:
• spaCy (advanced usage, not basic pipelines)
• Training custom NER models
• Data annotation and dataset preparation
• Model evaluation (precision, recall, F1)
• Strong Python skills for text processing
• Understanding of:
• Tokenization challenges
• Entity boundaries and labeling consistency
• Handling noisy / unstructured text

What You Will Do

• Design and train custom NER models using spaCy
• Prepare and clean training datasets (annotation, labeling standards)
• Evaluate and improve model performance iteratively
• Handle edge cases and ambiguous entity extraction
• Integrate NLP outputs into backend/data pipelines
• Optimize models for accuracy and performance

Strict Performance Rules

• Zero hand-holding. You are expected to understand the full NLP pipeline
• Deadlines are absolute. Miss once → warning. Repeat → termination
• Daily reporting required (metrics, experiments, results)
• Models must show measurable improvement, not random attempts
• You must justify decisions (labels, architecture, evaluation)

Disqualification Triggers

Immediate removal if:

• You cannot explain how your NER model was trained
• You rely only on pre-trained models without customization
• Your dataset is poorly labeled or inconsistent
• You cannot interpret evaluation metrics
• You disappear or fail to report progress

Selection Process (Strict Filtering)

• Project Review (must include custom NLP work)
• Technical Task:
• Train or improve a small NER model
• Live Discussion:
• Annotation strategy
• Error analysis
• Final acceptance

Most candidates will fail at step 1.

What You Get (If You Pass)

• Real experience building custom NLP systems
• Deep understanding of NER in production environments
• Strong portfolio with measurable NLP results
• Fast-track to a paid NLP / AI role

Application Requirements

Submit:

• GitHub with NLP / spaCy work
• Description of a custom NER project (dataset, approach, results)
• Example of labeled data you created
• Answer:

“How would you improve a weak NER model with low recall?”

Final Note

If you have not trained a custom NER model yourself, this role will not work for you.

Pay: E£1,000.00 - E£8,000.00 per month

Application Question(s):

• Are you familiar with Spacy?

Language:

• English (Required)

Work Location: Remote