Location: Bengaluru, India (On-site)

Employment type: Internship, 6 months (with possible conversion to full-time)

Eligibility: Final-year students or recent graduates (B.Tech / BE / M.Tech / MSc in CS, Data Science, AI/ML or related)

Reports to: Tech Lead (day to day with our AI-focused full-stack engineer)

Stipend: As per industry standards

About Insureka

Insureka builds digital insurance platforms for the Indonesian market. We cover customer web and mobile apps, partner and agent portals, admin dashboards, and the services behind quotes, policy issuance, claims, payments and KYC. Our platform runs on Python microservices on Google Cloud (GKE), with React/Next.js web apps and Flutter mobile apps.

About the role

You will help bring AI into real insurance workflows: reading and extracting data from documents, automating manual review, and building assistants for customers and internal teams. You will build prototypes, measure how well they work, and help turn the best ones into production features.

What you will do

- Build prototypes on LLM APIs (Anthropic Claude, OpenAI, Google Vertex AI / Gemini): prompt design, structured JSON output, tool/function calling
- Build OCR and document-extraction pipelines for identity, vehicle and insurance documents, and measure accuracy
- Build and evaluate RAG systems: chunking, embeddings, vector search, answer quality




- Create evaluation datasets and compare models and prompts on accuracy, cost and latency
- Wrap working prototypes in Python (FastAPI) APIs the product can use
- Document experiments, results and recommendations clearly

What we are looking for

- Strong Python, with pandas and NumPy
- ML fundamentals: supervised learning, evaluation metrics, overfitting, train/test splits
- At least one project using an LLM API or an ML model (scikit-learn, PyTorch or similar)
- Basic SQL and REST APIs
- Git, and a GitHub profile or project portfolio

Good to have

- LangChain, LlamaIndex or agent frameworks
- OCR and document AI: Tesseract, Google Document AI, PaddleOCR
- Vector databases: pgvector, Pinecone, FAISS, Chroma
- FastAPI, Docker, GCP
- Model fine-tuning, prompt evaluation or red-teaming
- Kaggle, research papers or AI/ML hackathon projects

What you will learn

- Building production AI features, not just notebooks
- LLM evaluation, cost control and safety in a regulated industry
- Python microservices and cloud deployment
- The insurance domain: underwriting, claims, KYC

You will do well here if you

- Think experimentally and measure before concluding
- Are curious, learn fast and own your work
- Can explain results to both technical and non-technical people
- Have valuable written and spoken English

📌 AI/ML Engineering Intern (Bengaluru)
🏢 insureka
📍 Bengaluru