Looking for a job

Farhan M

@farhanm

Full Stack Engineer at Nexturn building an enterprise RAG assistant with hybrid search, knowledge graphs, and source citations.

India

Message

At Nexturn, I built retrieval for an enterprise AI knowledge assistant that combines vector and keyword search, semantic reranking, and a knowledge graph. I also developed Outlook and Confluence ingestion connectors and permission-aware retrieval so answers include source citations and respect Microsoft Entra ID access lists.

Previously at Starlite Infotech, I built AI knowledge and financial automation platforms with React.js, Next.js, FastAPI, and LLMs. My work improved knowledge search time by 70% for 10K+ enterprise users, reduced manual document processing by 65%, and included LLM evaluation pipelines and cloud deployments.

Experience

Work history, roles, and key accomplishments

Current

Full Stack Engineer

Current

Nexturn

Jun 2026 - Present (4 months)

Built an enterprise AI knowledge assistant with RAG and knowledge graph, combining hybrid search and semantic reranking. Developed retrieval layer with Azure AI Search, enforced permission-aware access, and shipped via Azure DevOps.

React Next.js Python fastAPI Azure AI Search Azure OpenAI Azure Cosmos DB Gremlin Pydantic Docker Kubernetes Terraform Azure DevOps CI CD RAG Knowledge Graphs Reranking Reciprocal Rank Fusion Microsoft Entra ID

Full Stack Engineer

Starlite Infotech

Sep 2023 - Dec 2025 (2 years 3 months)

Architected scalable React.js and Next.js interfaces with FastAPI microservices, reducing knowledge search time by 70% for 10K+ users. Built production-grade AI knowledge assistant with RAG pipelines, improving query accuracy by 45%.

React Next.js NodeJS Python fastAPI TypeScript Docker Kubernetes AWS ECS Terraform GitHub Actions CI CD RAG LLM Evaluation Prompt Engineering Embeddings Semantic Retrieval

Software Engineer

Starlite Infotech

Sep 2021 - Aug 2023 (1 year 11 months)

Designed FastAPI backend services with asynchronous Node.js orchestration for financial automation, reducing manual document processing by 65%. Fine-tuned LLMs for invoice parsing and validation, improving data accuracy by 45%.

Python TypeScript fastAPI NodeJS Docker Kubernetes AWS Sagemaker AWS Lambda GitHub Actions Terraform LLM Prompt Engineering NLP Next.js CI CD

Associate Software Engineer

Starlite Infotech

Nov 2019 - Aug 2021 (1 year 9 months)

Built full-stack LLM-powered code generation and refactoring tools with React.js, Node.js, and Python, increasing developer productivity by 35%. Developed automated prompt engineering frameworks and evaluation harnesses, reducing production bugs by 28%.