Sahil Verma

Implemented Hybrid Search for document retrieval process using BM25 and Vector similarity.

Improved search performance by 30% by implementing RRF reranking and building benchmark suites.

Developed RAG based retrieval pipelines and digitizing up to 1 million documents annually.

Leveraged Agentic Patterns such as Reflection, HITL to drive up to 90% accuracy for data extraction.

Developed a generalized connector layer to enable integration with 6000+ applications.

Created classifiers using LLMs with up to 97% accuracy without training, saving 100+ hours on annotations.

Tech Exposure: RAG, Benchmarking, AI Agents, Flask, PgVector, LangChain, LangGraph, Grafana

Developed a cart system for the e-commerce app, driving a 60% increase in company revenue.

Developed a middleware system to reduce boilerplate, increasing developer productivity by 50%.

Designed and built features like Online Payments, BNPL, Checkout, Authentication with over 5000 daily users.

Led the codebase refactoring effort and code reviews, reducing tech debt by more than 70%.

Peer reviewed over 80k lines of Python code from over 600+ pull requests to significantly reduce tech debt.

Developed test automation suites to improve QA productivity by nearly 50%.

Tech Exposure: Django, Odoo, Postgres, Redis, Swagger, AWS, Postman