Personal details
Mitali S.
Technical skills
About
I am a Software Engineer with over six years of experience specializing in high-throughput backend applications, cloud infrastructure, and advanced AI engineering. I have expertise in multi-tenant architectures, Kubernetes-based microservices, and orchestrating zero-downtime distributed systems that handle millions of daily messages. My recent work includes engineering enterprise-grade multi-agent RAG platforms utilizing LangGraph and optimizing pgvector retrieval pipelines for sub-150ms latency. Moving forward, I aspire to pioneer the architecture of next-generation intelligent platforms, leveraging robust, highly scalable cloud infrastructure to drive complex enterprise automation and high-availability system performance.
Work Experience
• Led zero-downtime database switchover strategy for a high-throughput real-time payment processing application handling over 2 million messages daily, leveraging coordinated Kafka listener activation across clusters ensuring seamless availability during migrations and patching, saving $XM quarterly.
• Autonomous Research Agent System: Engineered a multi-agent RAG orchestration platform utilizing LangGraph's Send API to decompose complex queries into parallel search workers; implemented a pgvector pipeline (1536-dim embeddings) delivering <150ms cosine-similarity retrieval. Developed an LLM-as-judge evaluation framework tracking groundedness, hallucination, and citation accuracy, integrating it into GitHub Actions to automatically block PRs on score regression. Instrumented full node-level traces via LangSmith to optimize latency, cost, and failure rates.
• Architected an automated error-correction workflow for the South Africa market that slashed payment resolution time from 24+ hours to <5 minutes, reduced manual work by 85%, and prevented $xK annual revenue leakage.
• Enhanced and maintained Payment Clearing Connector application, implementing critical features that improved transaction processing reliability, writing comprehensive test suites, and resolving complex production issues for a system handling 10M+ daily payment messages with 99.999% uptime requirements.
• Conceptualized, designed, and led development of infrastructure control plane to streamline infrastructure data aggregation, empowering 50+ teams to manage infrastructure and cloud resources in a compliant manner.• Engineered a multi-tenant architecture using Next.js 13 and TypeScript to manage cross-chain metadata for digital/physical assets and decentralized asset storage (Cloudinary/IPFS) to build a high-throughput EVM-compatible payment integration layer for distributed storefronts.
• Led integration of Bixby AI assistant to third party services expanding the reach of Bixby ecosystem and significant upswing in user engagement.
• Proposed and led development of a cost prediction web app based on Google Cloud that helped with cost reduction recommendation and enabled cloud cost governance. This led to 30% savings in the resource expenses for the R&D division.
• Designed and implemented robust scalable Kubernetes based microservice architecture over multi-cloud Infrastructure for Bixby voice-controlled assistant components and successfully integrated with observability stack.
• Steered enhancement of CI strategy for early fault detection prior services rollout and moved to Canary deployments while improving the service availability by 20% margin.