Posted 1h ago
Loading this job…
Natural Language Processing (NLP) Intern
Posted 1h ago0 views
Parsing the messy middle of a job post
Job descriptions reach us as PDFs, pasted HTML, and occasionally as a screenshot someone retyped by hand. Your work over three months is to turn that mess into structured fields: role title, seniority, must have skills, nice to have skills, and location policy.
How the work runs
You will fine tune a small transformer from the Hugging Face hub on a labelled set of about twelve thousand Indian job posts, starting from a distilled BERT checkpoint and measuring it against a spaCy rule baseline that already exists. Expect to spend real time on the label set itself, because the first thing you will discover is that four annotators disagree about what counts as a required skill. Error analysis happens every Friday with your mentor. You ship the extractor behind an API endpoint and write the evaluation notebook the team keeps using after you leave.
What you should bring
Final year students and recent graduates are exactly who this is for. Bring Python, some exposure to tokenisation and embeddings from coursework or a side project, and the patience to read a confusion matrix twice before drawing a conclusion from it. The role is fully remote across India.
Responsibilities
- Clean and normalise job description text arriving as PDF, HTML and pasted plain text
- Extend and audit the annotation guidelines for skills, seniority and location policy
- Fine tune a distilled transformer from the Hugging Face hub on about twelve thousand labelled Indian job posts
- Compare model output against the existing spaCy rule based baseline
- Run weekly error analysis sessions and log the failure categories you uncover
- Wrap the final extractor in an API endpoint with tests
- Hand over an evaluation notebook the team can rerun without you
Requirements
- Final year student or recent graduate, no professional experience required
- Python fluency and familiarity with at least one deep learning framework
- Exposure to tokenisation, embeddings or transformer architectures from a course or project
- Comfortable reading precision, recall and confusion matrices
- Reliable internet and availability across Indian business hours
Skills
Benefits
- Paid monthly stipend with fully remote working
- Flexible hours arranged around your final semester timetable
- Weekly mentorship from the NLP lead
- Completion certificate and public credit in the release notes