Bikash Singhal

I’m a Staff-level Software Engineer and AI Engineer with 10+ years of experience building production software, backend systems, and AI applications. I’m currently building production-ready Agentic AI and RAG systems through Keystone Consultancy Services, with a focus on Python, LangGraph, Amazon Bedrock, FastAPI, PostgreSQL, Docker, and AWS.

My recent work includes Negotia, a multi-agent negotiation platform built with LangGraph and Amazon Bedrock. I designed stateful agent workflows for AI opponents, coaching, debriefs, strategy generation, and cross-session memory, along with FastAPI services, PostgreSQL persistence, automated testing, Docker containerization, and AWS deployment.

I also built a modular RAG platform with PDF ingestion, hybrid BM25 + FAISS retrieval, cross-encoder reranking, and Bedrock generation. Through performance optimization, I reduced cross-encoder startup time from 351 seconds to 73 seconds while maintaining retrieval quality, and achieved Recall@K of 1.000 on my evaluation dataset.

Previously, I was a Staff Engineer at Qualcomm, where I led architecture and delivery for QINTENT and worked with engineering teams across the United States and India. Earlier at Motorola, I worked on Android streaming and VOD SDKs, with hands-on experience in Java, C++, Android, testing, and customer integration.

I’m particularly interested in remote opportunities involving Agentic AI, LLM applications, RAG, AI evaluation, backend engineering, and cloud systems, where I can combine hands-on engineering with technical leadership and end-to-end ownership.