Open to opportunities

Nazina Nazar

@nazinanazar

Message

Machine Learning Engineer building scalable NLP, LLM, RAG, and AI systems with production-grade MLOps.

India

Message

What I'm looking for

I want to build production-grade ML/NLP/LLM systems—especially RAG, automation, and end-to-end ML pipelines. I’m excited to work on MLOps workflows: deployment, monitoring (Prometheus), CI/CD, and scalable services on AWS.

I’m a Machine Learning Engineer focused on building scalable ML, NLP, LLM, and AI systems using Python, FastAPI, PyTorch, and AWS. I care about delivering production-grade solutions end-to-end—from data pipelines to deployment and monitoring.

In my AI healthcare work, I built a production-grade multi-modal AI platform supporting text, voice, document OCR, and image-based diagnostics. I developed skin cancer and chest X-ray classification models (0.91 ROC-AUC with 83% recall for skin cancer; 0.71 ROC-AUC for X-ray), and engineered a real-time voice pipeline that reduced latency to 2.5–3.5s while supporting 50+ concurrent users.

I also build reliable RAG and automation systems. I created a Retrieval-Augmented Generation interview assistant using FAISS + Gemini embeddings with chunking and semantic search via LangChain LCEL, along with a FastAPI backend, rate limiting, and Prometheus monitoring—plus a Streamlit chat UI and CI/CD for quality and testing.

Beyond that, I develop privacy-first and scalable MLOps workflows, including JWT authentication with Argon2 hashing, MLflow experiment tracking, Docker-based multi-service deployments, and Kubernetes orchestration. I’m motivated to keep improving model performance, system reliability, and developer-friendly deployment practices.

Experience

Work history, roles, and key accomplishments

RAG Interview Assistant

AI Interview Assistant

Developed a Retrieval-Augmented Generation (RAG) system that answers interview questions from indexed PDFs using FAISS and Gemini embeddings. Implemented a recursive chunking + semantic search pipeline, a FastAPI backend with rate limiting and Prometheus monitoring, and a Streamlit chat UI.

Python Faiss Gemini API fastAPI Semantic Search Vector Databases Docker Prometheus Streamlit

AI Healthcare Platform

Curamyn

Built a production-grade multi-modal AI healthcare platform supporting text, voice, OCR, and image diagnostics using LLM workflows, Tesseract OCR, and PyTorch transfer learning CNNs. Achieved 0.91 ROC-AUC (83% recall) for skin cancer detection and 0.71 ROC-AUC for chest X-ray classification, with real-time voice latency of 2.5–3.5s for 50+ concurrent users.

Python fastAPI PyTorch LLM Tesseract Whisper Transfer Learning AWS EC2 Docker

Retail Sales Intelligence

Retail Sales Intelligence Dashboard

Built an end-to-end BI pipeline with ETL workflows and star schema design, including RFM-based customer segmentation and 90-day sales forecasting. Delivered interactive Power BI dashboards for KPI tracking and forecast visualization.

Python PostgreSQL Power BI ETL Star Schemas Prophet K Means Pandas NumPy

Customer Support Automation

Intelligent Customer Support Automation

Fine-tuned DistilBERT models for sentiment, severity, and intent classification and integrated them with a Gemini LLM for contextual responses. Built a FastAPI + Streamlit system with S3-based model auto-loading and Kubernetes-deployed, multi-stage Docker services.

fastAPI DistilBERT Gemini API Streamlit AWS S3 Kubernetes Docker GitHub Actions

E-Commerce Real-Time Analytics

Implemented a real-time analytics pipeline streaming synthetic e-commerce orders via Kafka, using Spark MLlib to predict prices with RandomForest and storing results in PostgreSQL. Built interactive Plotly Dash dashboards to visualize streaming trends.

Apache Kafka Spark Structured Streaming MLlib Random Forest PostgreSQL Plotly ETL Python

AI Resume Screening System

Developed an MLOps-based resume screening system using TF-IDF role classification and semantic similarity ranking with Sentence Transformers. Deployed with Docker, Streamlit, CI/CD, and Prometheus monitoring for production-style observability.

fastAPI scikit learn Sentence Transformers Docker Streamlit CI CD Prometheus TF IDF MLOps

Education

Degrees, certifications, and relevant coursework

Brototype Kerala

Data Science & Machine Learning Program, Data Science and Machine Learning

2024 - 2026

Data Science and Machine Learning Program at Brototype Kerala from 2024 to 2026.

UKF College of Engineering and Technology

Bachelor of Technology in Computer Science and Engineering, Computer Science and Engineering

2020 - 2024

Bachelor of Technology in Computer Science and Engineering at UKF College of Engineering and Technology (Kollam, Kerala) from 2020 to 2024, affiliated with APJ Abdul Kalam Technological University.