Headline

Senior Software Engineer @ Oracle | Building AI Powered Scalable Distributed Systems | Backend | LangChain | MCP | RAG | Multi Agent | Gold Medalist @NIT Raipur’23

About

I'm a Senior Software Engineer at Oracle, specializing in Agentic AI, Generative AI, distributed systems, and real-time, cloud-native applications. Over the past 3+ years, I've worked on building and scaling enterprise-grade systems, including an LLM-powered contact center AI platform deployed across multiple Oracle business units and serving 7,000+ agents. My work spans backend architecture, LLM orchestration, real-time audio and event streaming, frontend development, and cloud infrastructure. Some highlights of my work: * Designed and built core microservices and LLM orchestration workflows for enterprise GenAI applications. * Reduced GenAI summarization API calls by ~99% through feature redesign. * Improved real-time transcript processing throughput by 10×. * Built distributed audio streaming and transcription pipelines supporting 35+ languages. * Delivered multi-region, production-ready services using Kafka, WebSockets, Kubernetes, and Oracle Cloud Infrastructure. * Developed agentic automation workflows for IT issue diagnosis, resolution, and intelligent escalation. I'm particularly interested in building intelligent, autonomous systems that combine LLM reasoning, tool integration, reliable backend engineering, and scalable infrastructure to solve real-world problems. I'm building my agentic AI portfolio. My latest project is an autonomous incident response agent in LangGraph — it diagnoses production alerts, runs self-correction loops when confidence is low, and generates structured post-mortems without human input. I'm working with LangChain, CrewAI, and MCP to go deeper on multi-agent coordination, RAG pipelines, and tool-use patterns. Areas of interest: Agentic AI | LLM Applications | AI Agents | Backend Engineering | Distributed Systems | Real-Time Streaming | Cloud-Native Architecture My edge: I understand what production-grade AI systems actually require — not just the model, but the orchestration, the observability, and the failure modes. That's the gap I'm closing, and the role I'm building toward.

Location

India

AI stack

claude code

codex

hugging face

langchain

langgraph

OpenAI Playground

Work experience

Senior Software Engineer at Oracle, Oct 2025 - Present · 1 yr 2 mos, Bengaluru, Karnataka, India, Oracle's GenAI-powered contact center platform, a Top-10 Global Oracle initiative live across multiple business units. Team recognised with the Oracle Summit Award FY25 Q4. • Designed, deployed and owned the central Node.js LLM orchestration service, integrating 8+ microservices through REST APIs, Kafka and WebSockets for a production Agent Assist platform serving 7k+ agents. • Designed and built Python/Fast API microservices using LangChain for real-time RAG-powered Agent Assist, supporting across 35+ languages. • Owned the user-facing conversational UI delivering real-time RAG suggestions and post-call LLM summaries; contributed to lower call handling time through actionable, contextual assistance. • Redesigned the LLM summarization workflow to trigger once after a call, cutting LLM API requests ~99%, lowering operating costs and reducing redundant orchestration. • Optimized a scalable real-time transcript pipeline processing transcripts across 150–200 concurrent sessions; achieved 10× throughput and 90% lower per-message processing latency (~50 ms to ~5 ms). • Engineered production voice streaming, STT and recording using SBC/SIPREC, WebSockets, Kafka and cloud storage; integrated WeNet, OpenAI Whisper and OCI Speech-to-Text and optimized concurrent audio handling. • Deployed and maintained services across regions with Docker/Kubernetes (OKE); diagnosed production incidents using OCI Monitoring/Grafana, performed root-cause analysis and implemented reliability fixes. Led secure practices Oauth2, vault based secrets, and SSO integration. Improved scalability through non conventional approaches. Authored runbooks, technical documentations and mentored junior engineers. Leveraged AI tools like codex for 40% faster development cycles. • Built a Slack-based L1 support agent with OpenAI Workspace Agents for FAQ and troubleshooting workflows, reducing manually handled L1 tickets.

Member Technical Staff at Oracle, Jun 2023 - Oct 2025 · 2 yrs 5 mos, Bengaluru, Karnataka, India, Joined Oracle from campus as part of the contact center engineering team. Contributed to the migration of Oracle's contact center platform and was moved onto the GenAI contact center initiative within the first few months — one of Oracle's Top-10 Global GenAI projects. Contributed to contact center platform migration — understanding legacy architecture and helping transition components to the new stack. Ramped quickly onto the GenAI contact center platform team, taking on frontend ownership and key backend initiatives early in tenure. Built foundational knowledge of Oracle's cloud-native infrastructure: OKE, Docker, Kubernetes, WebSockets, and microservices architecture. Collaborated across engineering, product, and QA in an Agile environment — contributed to design reviews, code reviews, and onboarding of team processes.

Education

Bachelor of Technology, Computer Engineering, National Institute of Technology Raipur

Intermediate, DPS Bulandshahr

Public links

github: https://github.com/itzyash01

leetcode: https://leetcode.com/itzyash_01