AI Engineer · AWS Certified · 3★ CodeChef
Sai Sarthak Mohapatra.
Building LLM-powered products and distributed systems at the intersection of AI and enterprise SaaS.
I'm an AI Engineer with 5+ years of experience building distributed systems, cloud-native SaaS platforms, and LLM-powered products. Currently at Showpad, I'm architecting the Deal Intelligence platform — unifying CRM, Gong, Zoom, and Teams data into deal graphs that deliver real-time AI insights and in-call assistance to enterprise sales teams. I specialize in multi-agent LLM orchestration, RAG pipelines, AWS serverless infrastructure, React, TypeScript, Node.js, and GraphQL. Previously at SysCloud, I led backend architecture for enterprise backup systems serving 10K+ global users, built GenAI-assisted diagnostics with Claude APIs, and engineered secure OAuth 2.0 authentication workflows. I'm AWS Certified and a 3-star competitive programmer on CodeChef. I don't just write code — I architect systems that scale invisibly and deliver experiences people genuinely enjoy using.
Expertise
Technologies I work with.
Carefully chosen tools and frameworks that let me move fast without compromising quality.
Selected Work
Things I've built.
Each project represents a unique challenge — from AI-powered platforms to distributed orchestration engines.
Backup Jobs 3.0
Distributed job-orchestration engine built with Node.js, GraphQL, PostgreSQL and AWS. Improved platform reliability by 40% and enabled AI-assisted failure diagnostics using Claude-based LLM pipelines for 10K+ enterprise users.
Product Design Framework
Scalable React component-driven architecture that reduced frontend code changes by 70%. Includes JSON-driven UI enabling runtime interface configuration without redeployment, improving iteration velocity by 40%.
Enterprise Security & Compliance Suite
Modular backend architecture supporting Ransomware Detection, Compliance Monitoring, eDiscovery, Archival and Data Insights. Secure OAuth 2.0 with PKCE and AWS Lambda authorizers.
MeetArc — AI Meeting Intelligence Platform
Full-stack AI-powered meeting intelligence platform built across three microservices. Next.js 16 frontend with React 19 Compiler, MUI v7, TanStack Query v5, and NextAuth/Supabase auth. Offline-first IndexedDB upload queue with exponential backoff and cross-session recovery. Express.js backend with Prisma/Neon PostgreSQL, BullMQ/Redis job processing, and dual AI integrations (Claude + OpenAI). Autonomous meeting bot using rebrowser-playwright for stealth Google Meet recording with state-machine lifecycle, three-layer participant detection, grace period auto-leave, and two-phase Supabase Storage upload with crash recovery. Multi-cloud storage (AWS S3, Azure Blob, Supabase), Prometheus metrics, and fully containerized with Docker Compose.
ScreenAppBot — MeetArc
Universal meeting bot to record Google Meet, Zoom, and Microsoft Teams with a single API. Production-ready, free to use, extend, and scale.
View on GitHubFood Coup Mania
Full-stack React app with GraphQL (Apollo Client/Server), Hasura, and PostgreSQL. Features cuisine-based filtering, user auth, and restaurant offer discovery deployed on Heroku.
View on GitHubFoodInesta Restaurant App
React restaurant application with unit testing via Jest, API integrations with Axios, and responsive UI with dynamic menu rendering.
View on GitHubPlane Shooter 2D Game
Android 2D animated game built in Java. Shoot missiles at planes using a tank, track scores, health bar and best scores. Uses Bitmap Factory, Canvas OnDraw, and OOP concepts.
View on GitHubQuiz Game — Python
Automated quiz platform with difficulty levels (Easy, Medium, Tough), user registration, instant scoring, and a fully automated evaluation system.
View on GitHubLeetCode & CodeChef Practice
Comprehensive competitive programming solutions from beginner to hard — a journey from scratch to 3★ CodeChef with well-documented approaches.
View on GitHubExperience
Where I've worked.
Building scalable platforms and leading cross-functional delivery at enterprise scale.
Building AI-powered deal intelligence infrastructure at Showpad, a leading sales enablement platform — unifying revenue signals across CRM, communication, and conversation tools to deliver real-time insights at the point of sale.
- ▸Architecting Showpad's Deal Intelligence platform — building a multi-source ingestion pipeline that unifies CRM data (Salesforce, HubSpot), meeting transcripts (Gong, Zoom), and collaboration signals (Microsoft Teams) into a normalized event stream using AWS EventBridge, SQS, and Lambda
- ▸Designing deal graph data models — representing accounts, contacts, activities, and deal stages as nodes/edges to enable traversal-based risk scoring, stakeholder influence mapping, and engagement gap detection
- ▸Building real-time AI assistance during live sales calls — LLM inference pipelines with streaming output, RAG over deal history and CRM context, and tool-use agents that surface battle cards, pricing objections, and next-best-action recommendations mid-conversation
- ▸Engineering LLM orchestration layers using multi-agent architectures (planner + specialist agents) that synthesize signals from Gong conversation intelligence, CRM activity logs, and calendar metadata into structured deal narratives and forecast confidence scores
- ▸Developing AI insight APIs consumed by Showpad's frontend — webhook-driven deal health updates, anomaly alerts (deal slippage, champion change, competitor mentions), and conversational analytics dashboards
- ▸Integrating with Showpad Genie Agent Studio and MCP server infrastructure to expose deal intelligence as governed, queryable context for downstream AI agents across the Revenue Effectiveness Operating System
Leading architecture and development of distributed orchestration systems, AI-powered diagnostics, and multi-region enterprise SaaS infrastructure.
- ▸Architected Backup Jobs 3.0 — a distributed job-orchestration engine built with Node.js, GraphQL, PostgreSQL and AWS, improving platform scalability and reliability by 40%
- ▸Designed modular backend architecture supporting 5 enterprise capabilities: Ransomware Detection, Compliance Monitoring, eDiscovery, Archival, and Data Insights — serving 10K+ global users
- ▸Enhanced PostGraphile server internals and resolver pipelines for fine-grained orchestration control, optimizing execution flow by 25%
- ▸Built GenAI-assisted operational diagnostics using Claude LLM APIs — enabling automated failure analysis, contextual debugging insights, and intelligent remediation suggestions
- ▸Integrated LLM-powered sub-agents for contextual insights, anomaly summarization, and MCP orchestration workflows
- ▸Directed cross-region deployments across US & Tokyo data centers with automated multi-language workflows and GenAI-assisted validation pipelines, reducing deployment overhead by 35%
- ▸Drove end-to-end feature delivery as cross-functional lead — shaping architecture, AI integration strategy, and engineering best practices across teams
Built scalable frontend architecture, secure authentication systems, and high-throughput GraphQL APIs for enterprise SaaS backup products.
- ▸Engineered a scalable Product Design Framework for React-based applications — component-driven architecture that reduced frontend code changes by 70% and scaled apps by 50%
- ▸Developed JSON-driven UI architecture allowing runtime interface configuration without redeployment, improving product iteration velocity by 40%
- ▸Implemented secure authentication workflows using OAuth 2.0 with PKCE and custom AWS Lambda authorizers, reducing authentication-related issues by 30%
- ▸Designed and optimized GraphQL APIs backed by PostgreSQL — enabling high-throughput backend services and efficient query performance across large-scale data workflows
- ▸Integrated AI-assisted component generation, design validation workflows, and LLM-powered configuration assistants to accelerate delivery
- ▸Crafted pixel-perfect user interfaces for SaaS backup applications using React, TypeScript, Material UI, and CSS — translating complex requirements into intuitive experiences
Let's build something great.
Open to new opportunities. Have an idea or just want to say hello — my inbox is always open.