Generative AI Intern

Our Company:

Claw LegalTech

Location:

Duration:

Stipend:

PPO Opportunity:

Role Overview

AI/ML Research Intern

retrieval & reranking pipeline

Responsibilities

• Candidate Generation:

  Build smarter hybrid retrievers combining vector search, keyword search, section-aware lookups, and numeric anchors. Balance recall vs. noise in multi-method retrieval.

• Deduplication & Fusion:

  Move beyond “first-hit wins” deduplication, design better aggregation across multiple retrievers.

• Global Reranking:

  Experiment with reranking strategies (cross-encoders, weighted averages, calibrated heuristics). Find stable scoring methods across multi-query batches.

• Scoring Heuristics:

  Replace brittle hand-coded boosts with feature-based reranker signals, tuning weights with data.

• Cutoff Strategies:

  Design dynamic thresholds (score-based) instead of rigid top-3 returns, ensuring robust context passing to downstream LLMs.

Requirements

• Past projects with

  vector databases

  (FAISS, Pinecone, Weaviate, Milvus, Qdrant, or similar).
• Strong understanding of

  RAG pipelines

  — from indexing and chunking to reranking and evaluation.
• Experience in

  prompt engineering

  for retrieval/answerability tasks.
• Comfort with Python, HuggingFace, OpenAI/Cohere APIs, and evaluation metrics (nDCG, Recall@k, MRR).
• A mindset to

  diagnose, tune, and iterate,

  not just use defaults.

What will you gain

• Work directly on

  production-grade legal AI

  , solving problems where accuracy is critical.
• Collaborate with a technical team that’s already scaled AI products to 100k+ users.
• Exposure to cutting-edge techniques in

  hybrid retrieval, reranking, and multi-agent reasoning

  .
• A culture that values research-driven solutions, rapid experimentation, and real-world impact.

Perks and Benefits

• Fixed stipend of

  ₹8,000 per month

  .
• Performance-based incentive of up to

  ₹2,000 per month

  .

• Work from anywhere (remote role)

  .

• PPO (Pre-Placement Offer)

  available upon successful completion of internship.