Utkarsh Rana
Verified Expert in Engineering
Full-stack and Distributed Systems Engineer and Developer
Dehradun, India
Toptal member since October 22, 2026
Utkarsh is a full-stack and distributed systems engineer with 4 years of experience. As a co-founder and CTO of Zittle, he built a multi-tenant SaaS platform on serverless AWS that serves 6,000+ users and features AI booking agents and MCP servers for ChatGPT. At Flutter International, Utkarsh built Flink stream processors that handle 10+ million daily events and cut a DynamoDB table's cost by 93% (around $580,000 per year). He excels at AWS, TypeScript, Kafka, and LLM-powered products.
Portfolio
Experience
- Java - 4 years
- Amazon Web Services (AWS) - 4 years
- React - 3 years
- TypeScript - 3 years
- Apache Flink - 2 years
- MongoDB - 2 years
- Next.js - 2 years
- Apache Kafka - 2 years
Work Experience
Co-founder & CTO
Zittle Technologies
- Built and scaled a multi-tenant SaaS booking platform from 0 to 1 on serverless AWS that lets businesses launch branded booking sites in minutes, reaching 6,000+ users, 7,000+ bookings, and 100+ onboarded businesses (10-15 active).
- Engineered overbooking-proof seat holds, including atomic conditional increments, MongoDB transactions, and idempotent expiry, and enforced tenant isolation on every query, securing all 132 API routes against forged host headers.
- Deployed GPT function-calling booking agents on WhatsApp, Instagram, and ChatGPT Apps as a Meta Tech Provider, integrating Razorpay payments and calendar sync so each business sells across web and chat channels.
- Built 2 MCP servers for ChatGPT Apps: a multi-tenant catalog server with interactive widgets and a Shopify commerce app implementing MCP OAuth end-to-end, including RFC 9728/8414, dynamic client registration, and refresh tokens.
- Built an AI document-intelligence and revenue reconciliation platform that extracts financial data from PDFs, emails, spreadsheets, PMS reports, and OTA settlements using OCR and LLM agents, flagging only mismatches for review.
- Reduced storefront hosting costs by approximately 85% ($34 to under $5 per month) through tag-based Next.js cache invalidation, and halved Lambda cold starts (0.8-1.6s to 0.4-1.0s) by dividing a 100 MB bundle into 4 sub-10 MB functions.
- Led engineering and product strategy for 3+ engineers, owning architecture, code review, and CI standards, including lint, type-check, build, and dependency audit on every PR, while iterating the product with customers and accelerator programs.
Big Data Engineer
Flutter Entertainment
- Built 10+ production Apache Flink stream processors on Kafka and AWS, handling 10+ million real-time events daily, powering recommendation engines, behavioral analytics, and customer intelligence for PokerStars.
- Enabled rapid A/B experimentation for data science by building a configurable post-processing framework for the RELIC recommendation engine, so ranking-strategy changes no longer required code deployments.
- Saved about $580,000 annually (93%) on a DynamoDB table during the on-prem to AWS migration by consolidating around 300 attributes into 4 Kryo-serialized groups, reducing item size from 6.5 KB to under 1 KB, and cutting writes from 7 WCU to 1.
- Reduced Flink compute costs by 60% ($145 per day to $58 per day across 6 processors, about $32,000 per year) by implementing a custom autoscaler driven by real-time workload metrics, replacing AWS Managed Flink's conservative downscaling policy.
- Accelerated new business-event delivery by 70% with a Python code generator adopted team-wide across 1,000+ event types, and reduced integration-testing effort by 30% with a configurable Kafka mock-message generator.
- Eliminated manual game-metadata syncs and their redeploys with a CDC-driven Flink processor that streamed game updates from transactional systems into DynamoDB for downstream recommendation services.
Software Development Engineer
Amazon
- Developed features for CloudSpin, an event-driven AWS capacity platform using SNS, SQS, Lambda, and Step Functions. It was adopted by 30,000+ Amazon services to size fleets based on per-service scaling formulas for high-traffic events.
- Designed and built an end-to-end audit trail for service configuration, cutting incident investigation time by 90% by showing who changed what and when. Shipped bulk capacity overrides that cut correction effort by 90%.
- Unblocked adoption of automated stress testing by fixing a validation failure that broke about 30% of runs on NetScaler load balancers, picking the latest health record by timestamp, and adding bounded retries.
- Redesigned the service onboarding UI and added multi-scaling-formula support, cutting average onboarding time by 70% and removing per-service manual edits for teams onboarding entire fleets.
- Developed JAX-RS APIs and a React/Redux UI for FLO, a tool that stress-tests hosts with real production traffic to find their breakpoint TPM and right-size fleets.
- Identified an environment-isolation flaw where a non-production proxy path could reach production resources, and escalated it as an authorization boundary defect with candidate fixes.
Software Engineer Intern
Amazon
- Developed a near real-time notification system using SES, Lambda, Step Functions, Java, TypeScript, and React for a capacity platform supporting 30,000+ services, alerting owners via Slack, email, and tickets, reducing status-chasing by 80%.
- Automated monthly performance reporting with Step Functions and an EventBridge-scheduled API, eliminating 100% of the manual effort and the human errors that came with it.
- Reduced web page load time by 90% by optimizing React code and removing redundant API calls, which also decreased load on the back-end services.
- Extended integration tests in the admin service's CI/CD pipeline, preventing faulty releases from reaching production.
Project History
Zittle | Multi-tenant SaaS Booking Platform
https://zittle.inAs a co-founder and CTO, I designed and built the platform end to end and now lead a team of 3+ engineers. I built it on serverless AWS (Lambda, API Gateway, and CloudFront) with Next.js storefronts and MongoDB, provisioning a custom domain and TLS for each tenant at onboarding.
I engineered key features, including:
• Seat holds that prevent overbooking under concurrent load using atomic conditional updates in MongoDB transactions
• Tenant isolation on every query with a request authorizer across 132 API routes
• Razorpay payments with idempotent webhook handling
• GPT-powered booking agents on WhatsApp and Instagram as a Meta Tech Provider.
I also cut storefront hosting costs by 85% with tag-based cache invalidation and halved Lambda cold starts by splitting a 100 MB bundle into 4 smaller functions.
Education
Bachelor's Degree in Computer Science and Engineering
DIT University - Dehradun, Uttarakhand, India
Certifications
Learning Spring with Spring Boot
Infosys Certified Software Programmer
Infosys
Deep Learning
Coursera
Machine Learning with Python
Cognitive Class
Python Core and Advance
Udemy
Skills
Libraries/APIs
React, REST APIs, Node.js, WhatsApp API, Instagram API, OpenAI API, Claude API, Lombok, API Development, Back-end APIs
Tools
Git, Claude, Amazon CloudFront CDN, Mongoose, Amazon CloudWatch, AWS Cloud Development Kit (CDK), AWS Step Functions, Terraform, Claude Code, Prisma, AWS CloudFormation, Amazon Simple Queue Service (SQS), Facebook Business Manager, Amazon Simple Email Service (SES), Amazon Simple Notification Service (SNS), Observability Tools, Codex, Amazon Athena, Figma
Languages
TypeScript, Java, JavaScript, Python, SQL, CSS, HTML, C#, Python 3
Paradigms
Microservices, Model Context Protocol (MCP), Serverless Architecture, Object-oriented Programming (OOP), Agile Software Development, Microservices Architecture, REST, Event-driven Architecture, Event-driven Programming, API Architecture, Search Engine Optimization (SEO)
Platforms
AWS Lambda, Amazon Web Services (AWS), Apache Flink, Apache Kafka, Databricks, Google Cloud Platform (GCP)
Frameworks
Serverless Framework, SST, Next.js, Express.js, Dagger 2, Redux, Spark, Google Guava, Tailwind CSS, Material UI, Spring Boot
Storage
MongoDB, PostgreSQL, Amazon DynamoDB, Amazon S3 (AWS S3), Database Management Systems (DBMS), NoSQL, Databases, Google Cloud
Other
Full-stack Development, Full-stack, APIs, API Integration, API Design, Software Architecture, Back-end, Software Engineering, OpenAI GPT, Amazon EventBridge, CI/CD Pipelines, Computer Science, Data Structures, Algorithms, AI Agents, AWS Secrets Manager, AWS Managed Service for Flink, Artificial Intelligence (AI), Back-end Development, Software Development Lifecycle (SDLC), Software Development, Front-end Development, Email Integration, Payments, Responsive UI, React Hooks, Agentic AI, Architecture, Enterprise Architecture, Solution Architecture, AI Tools, Data Engineering, GitHub Actions, Tailwind UI, Technical Leadership, Team Leadership, Front-end, Large Language Models (LLMs), Bug Fixes, AI-assisted Development, Multi-tenant SaaS, SaaS, Cloud, Agentic Coding, Web Development, RESTFul APIs, Dashboards, Security, Smithy, AI Development, AI Integration, ChatGPT API, User Interface (UI), Rest APi, Webhooks, API Testing, Computer Networking, Operating Systems, Amazon API Gateway, Razorpay, AWS Managed Service for Kafka, Change Data Capture, Vite, Supabase, Leadership, Deep Learning, Machine Learning, Convolutional Neural Networks (CNNs), Application Security, Progressive Web Applications (PWAs), Distributed Systems, Performance Optimization, Multithreading, Data, Scaling, Concurrency, User Experience (UX), Front-end Design
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring