Prakash Thakuri Senior Software Engineer · Tech Lead

Software that holds up in production — at Fortune 50 scale and startup speed.

  • Full-stack systems
  • Distributed, cloud-native architecture
  • Platform engineering
  • Production AI
  • 8+years of production software, startup to Fortune 50
  • 60K+users migrated in a platform modernization I led end to end
  • Tech Leadfor platform services — authentication, identity, infrastructure, on-call
  • Production AILLM features, RAG pipelines, and MCP-powered agents in a live product
Experience across
  • ACE Virtual Shooting
  • ABC Fitness
  • The Home Depot
  • Fiserv
  • Ambit Techno

Selected engineering work

Systems I’ve owned end to end.

Three pieces of production work from tech-leading a startup platform-services team — each one spans architecture, implementation, infrastructure, and the on-call that follows.

  1. ACE Virtual Shooting · Platform modernization

    Migrating a 60K+ user platform off no-code.

    Context
    The customer-facing platform, serving 60K+ users, ran on Bubble.io — a no-code builder. The goal was to move it onto services the team owns, with real authentication and infrastructure-as-code behind it.
    My ownership
    Architecture, execution, and rollout — the service architecture, authentication, and the AWS infrastructure rollout — as tech lead for platform services.
    Engineering approach
    • Replaced Bubble.io with NestJS (TypeScript) services; platform services — authentication, identity, shared infrastructure — ran serverless on AWS Lambda, Cognito, DynamoDB, and RDS PostgreSQL.
    • Defined the whole AWS estate as Terraform — Lambda, DynamoDB, RDS, Cognito, S3, CloudWatch, WAF, SES — with GitHub Actions CI/CD and Grafana/Prometheus observability.
    • Consolidated email into a Lambda-based template system and sunset SendGrid, reducing vendor surface area and recurring spend.
    • Set up a client-side monorepo for the admin and customer portals.
    Outcome
    • 60K+ users migrated to the new platform.
    • Every piece of AWS infrastructure defined in Terraform and watched in Grafana; SendGrid retired.
    Technologies
    NestJS · TypeScript · AWS Lambda · Cognito · DynamoDB · RDS PostgreSQL · Terraform · GitHub Actions · Grafana
    System map
    1. Clients
      • Customer portal
      • Admin portal
      • VR & mobile apps
    2. Services
      • NestJS services
      • Magic-link auth (Lambda + Cognito)
      • Email templates (Lambda)
    3. Data
      • DynamoDB
      • RDS PostgreSQL
      • S3
    4. Platform
      • Terraform
      • GitHub Actions CI/CD
      • Grafana + Prometheus
      • CloudWatch
      • WAF
      • SES
  2. ACE Virtual Shooting · Platform services

    Magic-link authentication across web, VR, and mobile.

    Context
    A single sign-in system had to work across the web app, VR, and mobile deep-link flows — and it sat on the platform-services on-call rotation I carried as primary.
    My ownership
    Designed, hardened, and operated the authentication service end to end, including its observability.
    Engineering approach
    • Serverless magic-link flow on AWS Lambda, Cognito, and DynamoDB, extended to web, VR, and mobile deep links.
    • Built an auth-error observability pipeline from scratch (Terraform + DynamoDB + Grafana) tracking 400 and 500 errors across every sign-in flow.
    • Used that pipeline to detect a live WAF attack and mitigate it — hardening WAF rules and the RabbitMQ/DLQ retry logic in response.
    Outcome
    • One authentication path for web, VR, and mobile, with improved sign-in response time.
    • A live attack detected and mitigated through the observability built for it.
    Technologies
    AWS Lambda · Amazon Cognito · DynamoDB · Terraform · Grafana · AWS WAF · CloudWatch
    System map
    1. Sign-in surfaces
      • Web app
      • VR
      • Mobile deep links
    2. Magic-link auth
      • AWS Lambda
      • Cognito
      • DynamoDB
    3. Observability & defense
      • Auth errors → DynamoDB → Grafana
      • WAF rules
      • Retry logic (RabbitMQ / DLQ)
  3. ACE Virtual Shooting · Applied AI

    Production AI on Anthropic Claude — features, RAG, and MCP agents.

    Context
    Gen AI features inside a live customer product — owned end to end, from the model integration to the data they draw on.
    My ownership
    End to end: the customer-facing LLM capabilities, the RAG pipelines behind them, and the agent tooling used by engineering.
    Engineering approach
    • Customer-facing LLM capabilities built on Anthropic Claude.
    • RAG pipelines over internal data, so answers are grounded in the product’s own information.
    • AI agents powered by custom MCP (Model Context Protocol) servers, built to accelerate developer productivity.
    Outcome
    • LLM features live for customers.
    • Custom MCP servers and agents that accelerated developer productivity.
    Technologies
    Anthropic Claude · LLM integration · RAG pipelines · MCP servers · AI agents

How I engineer

How I engineer.

Enterprise discipline

Five companies taught me the same lesson five ways: production is the only environment that matters. I write code that survives compliance review, on-call, and the engineer who inherits it.

Startup velocity

On a startup platform team, shipping this week beats perfect next quarter. I scope ruthlessly, choose boring technology, and keep the product working while the platform matures.

Full-stack ownership

I don’t hand off at the API boundary. Schema, service, pipeline, interface, dashboard — owning the whole slice is how I keep systems coherent and postmortems short.

AI-augmented, human-reviewed

Agentic tools multiply output, not judgment. I build LLM products for users and use Claude Code daily myself — and keep design decisions, reviews, and merges firmly human.

Experience

Five companies. Five definitions of “production.”

A startup platform team, fitness SaaS at scale, Fortune 50 retail, fintech, and a full-stack start in Kathmandu — each one sharpened a different engineering muscle.

  1. ACE Virtual Shooting

    Senior Software Engineer / Tech Lead

    Sports Tech Startup

    Tech lead for the platform-services team — authentication, identity, shared infrastructure, and the product’s Gen AI features. Primary on-call for platform services.

    • Drove the end-to-end migration of the customer-facing platform from Bubble.io to NestJS for 60K+ users — architecture, execution, and rollout.
    • Designed and hardened serverless magic-link authentication on AWS Lambda, Cognito, and DynamoDB, spanning web, VR, and mobile deep-link flows.
    • Built Gen AI features end-to-end with Anthropic Claude: customer-facing LLM capabilities, RAG pipelines over internal data, and AI agents powered by custom MCP servers.
    • Built auth-error observability from scratch (Terraform + DynamoDB + Grafana), then used it to detect and mitigate a live WAF attack — hardening WAF rules and retry logic in response.
    • Owned the AWS estate as Terraform code — Lambda, DynamoDB, RDS PostgreSQL, Cognito, S3, WAF, SES — with Grafana/Prometheus observability and GitHub Actions CI/CD.

    Technologies: NestJS · TypeScript · AWS · Terraform · DynamoDB · Cognito · Grafana · Anthropic Claude · MCP

  2. ABC Fitness

    Senior Software Engineer

    Fitness SaaS

    Java/Spring Boot microservices at scale for a platform processing memberships and billing for thousands of gyms.

    • Built secure RESTful APIs with JWT and AWS Cognito; deployed Spring Boot services on Kubernetes (EKS) with GitHub Actions CI/CD.
    • Ran high-throughput workloads on RabbitMQ messaging and Redis caching; tuned schemas, indexes, and caching to harden APIs.
    • Carried on-call and led production incident response across distributed services; upgraded legacy Spring Boot services and remediated CVEs.

    Technologies: Java · Spring Boot · Kubernetes · AWS · PostgreSQL · RabbitMQ · Redis · GitHub Actions

  3. The Home Depot

    Software Engineer II

    Fortune 50 Retail

    Full-stack work on the appliance e-commerce experience — React components in front, Java services and GraphQL behind.

    • Built reusable React components with Redux and SCSS to internal design standards for a high-traffic retail surface.
    • Developed Java/Spring Boot REST APIs and integrated GraphQL services on GCP for product data.
    • Investigated and resolved traffic spikes and performance regressions across staging, beta, and production during peak retail loads.

    Technologies: React · Redux · Java · Spring Boot · GraphQL · GCP · SCSS

  4. Fiserv

    Software Engineer

    Fintech

    Internal financial-data applications and a platform modernization — from legacy JBoss to React, Spring Boot, and OpenShift.

    • Developed Node.js/Express and React applications used daily for financial data processing and calculations.
    • Migrated legacy applications to React + Java/Spring Boot and rebuilt deployment pipelines from JBoss to OpenShift on AWS.
    • Established CI/CD automation on Azure DevOps, built a reusable core-UI React library, and mentored junior developers.

    Technologies: React · Node.js · Express.js · Java · Spring Boot · OpenShift · Azure DevOps

  5. Ambit Techno

    Software Developer

    Full-Stack · Kathmandu

    Where it started: full-stack delivery across React, Angular, and Spring + Hibernate.

    • Built full-stack web applications with React, Angular, and JSP frontends over Spring/Hibernate REST APIs.
    • Modernized legacy codebases to current standards — cutting operating costs by ~35% — and wrote UNIX shell scripts for scheduled batch jobs.

    Technologies: React · Angular · Java · Spring · Hibernate · MySQL

Core technical expertise

Depth first, breadth second.

The stack I’ve shipped and supported in production — not every tool I’ve ever touched.

Languages
  • TypeScript
  • Java
  • JavaScript
  • SQL
Frontend
  • React
  • Next.js
  • React Native
  • Redux
Backend
  • NestJS
  • Node.js
  • Spring Boot
  • Express
  • REST
  • GraphQL
Data
  • PostgreSQL
  • DynamoDB
  • Redis
  • MongoDB
  • MySQL
Cloud & platform
  • AWS (Lambda, Cognito, DynamoDB, RDS, S3, WAF, SES)
  • Terraform
  • Kubernetes (EKS)
  • Docker
  • GitHub Actions
Reliability
  • Grafana
  • Prometheus
  • CloudWatch
  • On-call & incident response
AI engineering
  • Anthropic Claude
  • LLM integration
  • RAG pipelines
  • AI agents
  • MCP servers

Additional experienceGCP · Azure DevOps · OpenShift · Jenkins · RabbitMQ · OpenSearch · Django · Angular · Firebase · OpenAPI / Swagger · Jest · Mocha · Playwright · SCSS · Agile / Scrum

About

The short version.

I’m a senior software engineer and tech lead in Austin, TX, with 8+ years of production work across a startup platform team, fitness SaaS, Fortune 50 retail, and fintech.

The work I enjoy most is owning a hard problem end to end: understanding what the product actually needs, designing the architecture, building it, standing up the infrastructure, and carrying the on-call that proves it works.

I started in Kathmandu, hold an M.S. in Computer Science from Pace University, and have spent the years since building systems other engineers have to live with — which is the standard I hold my own code to.

Based in
Austin, TX
Education
M.S. Computer Science, Pace University
Environments
Startup · SaaS · Fortune 50 retail · Fintech
Open to
Senior full-stack, platform, and tech-lead roles

Contact

Let’s build something reliable.

Based in Austin, TX. Open to senior full-stack, platform, and tech-lead roles — email is the fastest way to reach me.

Need contract help instead? Consulting & freelance work