Ayush Srivastava — Senior Software Engineer

I build production agentic AI systems. At Synopsys, I shipped a multi-agent LLM platform (LangGraph, GPT-4, ReAct) handling 1000+ daily tasks at 99.5% uptime, with full eval discipline: golden test sets, per-agent latency/cost tracing via Langfuse, and prompt variant A/B comparison baked in from day one. Before that I built Codewhiz at Mercedes-Benz R&D — a RAG-based enterprise developer search platform with vector embeddings + semantic ranking that cut query resolution time by 30%. What I actually care about: → Evaluation harnesses that catch regressions before prod does → MCP server/client architecture (built from scratch — subprocess JSON-RPC, versioned tool schemas, 17 tools) → Multi-tenant agentic systems with real observability, not logging as an afterthought → Failure taxonomy: tool failure, reasoning failure, low-confidence, silent failure — I track all four Stack: LangGraph · LangChain · MCP · Claude · GPT-4 · RAG · Qdrant · Azure AI Search · Langfuse · FastAPI · Kubernetes · Terraform Open to senior AI Engineer / Agentic AI roles where the work is actually hard.

Stackforce AI infers this person is a Backend-heavy AI Engineer specializing in multi-agent systems and LLM technologies.

Location: Hyderabad, Telangana, India

Experience: 3 yrs 2 mos