Website:
thevotum.com
Job details:
About VotumVotum is building AI infrastructure for the legal and hiring ecosystem. We work on problems like case-law retrieval, document intelligence (OCR + NLP), automated assessments, and workflow automation for law firms, enterprises, and government bodies.
Our systems process large volumes of unstructured data — legal documents, resumes, invoices — and turn them into structured, actionable insights.
Role OverviewThis is not a “notebook-only” internship.
You will work on real production problems — improving model accuracy, building pipelines, and solving messy data challenges across legal-tech and HR-tech use cases.
What You’ll Work On- Build and improve NLP pipelines for legal documents (summarization, classification, extraction)
- Work on LLM-based systems (prompting, evaluation, fine-tuning workflows)
- Process OCR outputs and improve structured data extraction accuracy
- Analyze datasets like:
- Case laws
- GST/tax documents
- Candidate assessments and resumes
- Design evaluation frameworks (accuracy, recall, hallucination detection)
- Write production-grade data scripts and pipelines
- Work closely with backend engineers to integrate models into live systems
What We’re Looking For- Strong Python skills (Pandas, NumPy)
- Solid understanding of ML basics + statistics
- Comfortable working with messy, real-world data (not just clean datasets)
- Ability to think in terms of accuracy, edge cases, and failure modes
- Basic knowledge of SQL
Bonus (High Signal)- Experience with LLMs / prompt engineering / RAG systems
- Worked on NLP projects (even small ones)
- Familiarity with OCR challenges or document parsing
- Experience deploying models or working with APIs
- Contributions on GitHub / real shipped projects
What You’ll Gain- Direct exposure to production AI systems (not academic work)
- Understanding of how AI is applied in legal + enterprise workflows
- Ownership of meaningful problems (not side tasks)
- Fast learning curve working with a small, high-output team
Click on Apply to know more.