Skip to content

Senior AI Full Stack DeveloperKakinada, India

I build AI-powered products end to end.

Full-stack engineer with 3+ years shipping production systems: real-time interview platforms, LLM evaluation pipelines and multi-role SaaS, built on React, React Native, Node.js and MongoDB.

View ProjectsView Resume
Open to senior full-stack and AI engineering roles
Portrait of Hanumanthu Buddha
Experience
3+ years in production
Scale
10,000+ users · 99.9% uptime
Delivery
6 production apps led
Credential
MongoDB Certified Developer

01Engineering profile

Full-stack ownership, from schema to store listing.

I am a full-stack software engineer who builds AI-integrated, real-time products from the first schema to the deployment pipeline. Over the last three years I have shipped production applications on Node.js, React, React Native and MongoDB, serving 10,000+ users with 99.9% uptime.

My work sits where backend architecture meets applied AI: event-driven services on Redis and BullMQ, live Socket.IO sessions, JWT and role-based access for multi-tenant platforms, and LLM evaluation pipelines that turn a candidate's spoken answer into a score and specific, actionable feedback.

I am most useful on problems that need the whole stack to work together: a product idea that has to become an API, a data model, a mobile app and a reliable release process. Most recently that has been Vaktora, an AI interview-prep app now live on Google Play.

  • 01

    Product engineering

    React and React Native interfaces backed by Node.js REST services and MongoDB data models.

  • 02

    Real-time & async systems

    Socket.IO sessions, Redis caching and BullMQ job pipelines that keep APIs responsive under load.

  • 03

    Applied AI

    LLM evaluation pipelines, speech-to-text, model routing with fallbacks, and comparison of model outputs.

  • 04

    Delivery

    Docker, CI/CD and Google Play releases, with auth and role-based access designed in from the start.

03Selected work

More production systems.

Projects chosen for technical depth and real-world use. Where the code is private, the work is presented as a case study.

LSRW Communication Assessment Platform

AI-driven language proficiency assessment for institutions

Delivered · adopted by 3+ institutions

A modular platform that evaluates student proficiency across Listening, Speaking, Reading and Writing. AI grades spoken and written responses, subjective tasks fall back to hybrid human review, and educators get real-time reporting across 1,000+ assessments per session.

  • 20+ interactive testing modules: dictation, comprehension, storytelling, response selection and speaking tasks
  • Automated AI grading for speech and written responses, cutting manual evaluation time by 60%
  • Hybrid review path so subjective tasks keep a human in the loop
  • Test orchestration and analytics pipelines processing 1,000+ assessments per session

Role · Full Stack Developer

  • Next.js
  • Node.js
  • MongoDB
  • OpenAI API
Read case study
Delivered to a client. Source code and deployment are private, so no public link is available.

Village Finance

Multi-tenant lending and collections platform for village finance lines

In testing · backend live on Google Cloud

One codebase serving many lending businesses, each with its own admin login and fully isolated lines, collectors, customers and loans. An Expo mobile app for owners and field collectors, a Spring Boot API on Google Cloud Run with Neon PostgreSQL, and a React console for onboarding tenants.

  • Tenant isolation enforced from the signed JWT alone, so no request can name another client's data
  • Loans, installments, weekly book closing, capital, partners, expenses and chit-fund schedules in one data model of 20+ tenant-scoped tables
  • One-phone-per-collector device binding with an audit trail and admin-side reset
  • Presigned Cloudflare R2 photo uploads and a signed-URL cache tuned for village mobile connections

Role · Full Stack Developer

  • React Native (Expo)
  • TypeScript
  • Expo Router
  • React Query
  • Zustand
  • Java 17
  • Spring Boot 3
  • Spring Security (JWT)
  • PostgreSQL (Neon)
  • Hibernate / JPA
  • Cloudflare R2
  • Docker
  • Google Cloud Run
  • React + Vite
  • Jest
Read case study
Private product in internal testing ahead of its Play Store release. Source code is private.

04AI engineering

Beyond calling an API.

Integrating a model is the easy part. The work is making evaluation reliable, fast and consistent inside a product people use every day.

Basic API usage

  • Call a hosted model from the request handler
  • Return raw text to the client
  • One provider, no fallback
  • No way to check whether scores are consistent

AI application engineering

  • Queue evaluation jobs off the request path
  • Return structured scores, grades and feedback
  • Local model first, hosted model as fallback
  • Compare model outputs and keep humans in the loop where needed

Vaktora evaluation pipeline

  1. 01

    Candidate answers

    The app records a spoken answer to a role-specific question and uploads it with the session context.

  2. 02

    Interview engine

    The backend validates the session, stores the answer and enqueues an evaluation job on BullMQ.

  3. 03

    Transcription

    Whisper converts the recording to text so the evaluator works from a transcript.

  4. 04

    LLM evaluation

    A local Kimi model evaluates the answer first. If it fails or times out, the job is routed to OpenAI.

  5. 05

    Scoring

    The evaluation is turned into an overall score, a grade and per-category scores for the report.

  6. 06

    Personalized feedback

    Strengths and specific fixes are written back to the report and pushed to the app over Socket.IO.

Evaluation as a pipeline, not a request

Model calls run in BullMQ workers backed by Redis. The API acknowledges immediately, results stream back over Socket.IO, and a failed call is retried or routed to a fallback instead of being lost.

Model routing with fallback

Vaktora evaluates with a locally hosted Kimi model first and falls back to the OpenAI API when the local path fails or times out. The evaluation contract is identical on both paths.

Speech in the loop

Spoken answers are transcribed with Whisper before evaluation, so scoring works from a transcript and the report can show the candidate exactly what they said.

Comparing and checking model output

Outputs from both evaluation paths were compared on the same answers to keep scores consistent. On the LSRW platform, subjective tasks route to a hybrid human review rather than trusting the model alone.

05Technical expertise

Organized by the layer it runs in.

Only what I have used in production or shipped with. No percentage bars.

Frontend & Mobile

  • React.js
  • Next.js
  • React Native
  • Expo
  • TypeScript
  • JavaScript (ES6+)
  • Tailwind CSS
  • HTML5
  • CSS3

Backend & APIs

  • Node.js
  • Express.js
  • Java 17
  • Spring Boot 3
  • REST API design
  • WebSockets (Socket.IO)
  • FastAPI
  • JWT authentication
  • Role-based access control

Data & Caching

  • MongoDB
  • PostgreSQL (Neon)
  • MySQL
  • Redis
  • BullMQ
  • Schema design & indexing
  • Query optimization

AI Engineering

  • OpenAI API
  • Claude API
  • Whisper
  • Local LLMs (Kimi)
  • Prompt engineering
  • AI-based evaluation systems
  • Model comparison & fallback routing

Architecture

  • Distributed systems
  • Event-driven architecture
  • Asynchronous processing
  • Caching strategies
  • Performance optimization

DevOps & Delivery

  • Docker
  • CI/CD pipelines
  • Google Cloud Run
  • Cloudflare R2
  • Nginx
  • Linux
  • AWS
  • Azure
  • Git & GitHub
  • Postman
  • Google Play deployment

Languages

  • JavaScript
  • TypeScript
  • Java
  • Python
  • C++

06Experience

Where the work happened.

  1. Mar 2023 — Present

    Full Stack Developer

    Aditya University

    Own features end to end, from system design through CI/CD, on production platforms serving 10,000+ users.

    • Engineered full-stack applications on Node.js, React.js and MongoDB serving 10,000+ users with 99.9% uptime across production environments
    • Reduced system response latency by 40% by moving to event-driven backends with Redis and BullMQ for asynchronous job processing
    • Delivered real-time features on Socket.IO for live interview sessions, push notifications and multi-user workflows, supporting 500+ simultaneous connections
    • Improved API response times by 35% with Redis caching layers, cutting database query load
    • Secured multi-role platforms (Admin, Staff, Student) with JWT authentication and role-based access control
    • Refactored MongoDB schemas and introduced compound indexing, cutting average query execution time by 25%
    • Containerized applications with Docker and configured CI/CD pipelines for zero-downtime deployments across 3+ production services
    • Shipped 15+ major releases on schedule with product and design teams, and mentored 5 junior developers through code reviews and pairing, reducing PR review cycles by 30%
    • Node.js
    • React.js
    • React Native
    • MongoDB
    • Redis
    • BullMQ
    • Socket.IO
    • Docker
    • CI/CD

07Education & credentials

Background.

Education

Bachelor of Technology (B.Tech)

Electronics and Communication Engineering

Aditya Engineering College · 2019 – 2023

Certification

MongoDB Certified Developer, Associate

MongoDB, Inc. · 2025

Verify on Credly

Achievements

  • Mentored 700+ students in full-stack development (MERN stack, React Native); 85%+ completed capstone projects and secured developer roles.
  • Led development of 6 production-grade applications, all delivered on time with zero critical post-launch defects.
  • Achieved 40% application performance improvements across projects through backend optimization, caching and database design.

08Contact

Let’s build something useful.

Open to senior full-stack and AI engineering roles. If you are hiring for a team that ships AI-powered products, or you have a product that needs to go from idea to production, I would like to hear about it.

Based in Kakinada, India. Open to remote work.