Joel Henry Yellamelli

FullStack AI Engineer

Hyderabad, India 3+ yrs exp 77 · Very Good

About

No bio added yet

Skills & Expertise (33)

Python Advanced

8.5/10

Years Exp

React Advanced

8.5/10

Years Exp

JavaScript Advanced

8.0/10

Years Exp

Node.js Advanced

8.0/10

Years Exp

Machine Learning Advanced

8.0/10

Years Exp

Typescript Intermediate

7.5/10

Years Exp

AWS Intermediate

7.5/10

Years Exp

MongoDb Intermediate

7.0/10

Years Exp

Postgres Intermediate

7.0/10

Years Exp

Docker Intermediate

7.0/10

Years Exp

Kubernetes Intermediate

7.0/10

Years Exp

Flask Intermediate

7.0/10

Years Exp

GitHub Actions Terraform Cloud Infrastructure GCP Test-Driven Development FAISS Playwright Cypress ChromaDB CrewAI SageMaker MLflow LangGraph LangChain Tailwind CSS UI Shadcn FastAPI Svelte Next Go

Work Experience

FullStack Engineer

MONDEE

May 2026 - Present

Developed scalable full-stack applications using React, FastAPI, Python, MongoDB, Redis, TanStack Query, Zustand, React Hook Form, Zod, JWT, and Google/Microsoft SSO, following a clean architecture with asynchronous APIs and optimized database operations.

Software Engineer

ONE CONVERGENCE / DKUBE

Jun 2024 - Apr 2026

Architected and deployed 6 production microservices (Hybrid OCR, Embeddings, Segmentation, Classification, RAG Extraction) powering AI-driven document processing workflows using RayServe + Kubernetes for scalable model serving. Engineered a hybrid OCR + ML pipeline (Tesseract, EasyOCR, XGBoost) and built a RAG-based information extraction system leveraging ChromaDB for vector storage, improving classification and field-level extraction accuracy while reducing LLM token usage through structured context injection. Designed a Human-in-the-Loop RLHF retraining pipeline integrating NER-based feedback signals to continuously refine entity extraction models and improve downstream document classification accuracy.

Software Engineer Intern

ONE CONVERGENCE

Apr 2023 - Apr 2024

Contributed to an AI-powered greeting card platform, integrating image outputs from MidJourney and DALL·E into user-facing workflows. Provisioned and deployed services using Docker and AWS (EC2, S3).