Senior Software Development Engineer (SDE 3) | Backend & Agentic Systems

Job Description

WORK MODE: Hybrid (3 days WFO) LOCATION: Bangalore EXPERIENCE: 5+ YOE (minimum of 2–3 years in production LLM/agent systems) About the role This is a senior individual contributor role (~70% solo ownership, ~30% collaborative on high-blast-radius problems). You will own and extend production backend systems while also driving greenfield builds across the Recipe Engine, cloud infrastructure, and AI/agentic systems. You report directly to the Software Engineering Manager and join a backend team of four within a 12-person engineering org. Core responsibilities Design and architect complex features for autonomy, data logging, and reliability — HLDs and LLDs expected from day Own and extend the core backend stack powering autonomous operations at scale Build and maintain agentic workflows for recipe creation, feedback analysis, and internal tooling Build code-intelligence tooling for internal agents: AST-based chunking, LLM-driven graph extraction, hybrid retrieval (BM25 + graph + rerank) Improve retrieval accuracy through chunking, embeddings, and a benchmark-driven, eval-first approach Required skills Strong system design fundamentals: HLD, LLD, SOLID principles — tested directly in the interview loop Proven experience owning production LLM/agent systems (RAG pipelines, agent loops, tool calling/MCP) — not just API wrappers Hands-on with the retrieval data layer: vector databases, chunking strategies, embeddings, hybrid retrieval Experience tuning and improving existing RAG/retrieval pipelines; having built from scratch is a plus, not a gate Familiarity with AWS (primary stack) Nice to have Experience with knowledge graphs or graph-based retrieval (Neo4j, or equivalent) Hands-on experience with code-intelligence/retrieval frameworks (Cognee, LangChain, LlamaIndex, Haystack, or similar) Prior exposure to AST parsing or code-aware chunking strategies ABOUT THE COMPANY Our client is a fast-growing, Series A-funded robotics and AI company backed by a top-tier global VC. Their flagship product has received widespread recognition from leading technology publications and is already deployed at scale. They are building the next generation of autonomous systems powered by a sophisticated cloud backend and AI/agentic stack. ABOUT US Hyper Lychee Labs is an IT services and consulting company committed to delivering innovative and impactful solutions to a diverse client base, including Fortune 500 companies, the Government of India, and forward-thinking startups. With a focus on leveraging technology for transformation, the company provides custom software development, complex systems implementation, and strategic consulting. Our team of talented professionals collaborates to ensure exceptional results tailored to clients’ needs. Dedicated to fostering a culture of creativity, diversity, and inclusion, Hyper Lychee Labs strives to empower businesses to thrive in a rapidly evolving world.

• 5+ years of experience, with minimum 2-3 years in production LLM/agent systems
• Strong system design fundamentals: HLD, LLD, SOLID principles
• Proven experience owning production LLM/agent systems (RAG pipelines, agent loops, tool calling/MCP) — not just API wrappers
• Hands-on with the retrieval data layer: vector databases, chunking strategies, embeddings, hybrid retrieval
• Experience tuning and improving existing RAG/retrieval pipelines
• Familiarity with AWS (primary stack)
• Having built RAG/retrieval pipelines from scratch
• Experience with knowledge graphs or graph-based retrieval (Neo4j, or equivalent)
• Hands-on experience with code-intelligence/retrieval frameworks (Cognee, LangChain, LlamaIndex, Haystack, or similar)
• Prior exposure to AST parsing or code-aware chunking strategies