Meghna Nag — Software Engineer

About

Background

I build backend systems that move data reliably, scale under load, and simplify complex workflows.

My background spans AWS AppFlow, insurance platforms, and distributed systems, connecting messy data pipelines to production APIs that hold up under real load.

Languages

JavaPythonGoTypeScriptSQLScalaBash

Backend & APIs

Spring BootFastAPIgRPCRESTKafkaRabbitMQJPA

Cloud & Infra

AWS AppFlowAWS GlueS3LambdaGCPKubernetesDocker

Data & ML

PostgreSQLRedisSparkPyTorchAirflowMLflowSHAP

Work

Experience

Jan 2022 — May 2024

Persistent Systems

AWS AppFlow & Glue

Software Engineer

Built the AppFlow Connector SDK, cutting per-integration dev time by 66%
Built the ingestion framework behind 79 SaaS integrations, handling up to 100GB per flow into S3
Built the AWS Glue ETL framework with schema evolution, replacing one-off custom pipelines
Mentored 7 engineers; delivery velocity up 50% in 3 months

Dec 2016 — Dec 2021

Tata Consultancy Services

Solidarity Insurance / New India Assurance

Software Engineer

Dashboards went from 3 minutes to under 10 seconds using materialized views, stored procedures, and optimized SQL
Built async pipeline that generates 7 policy documents in under 10s using CompletableFuture
Automated premium recalculation, cutting manual errors by 50%
Migrated legacy EJB app to microservices (Spring MVC + AngularJS) with CI/CD

Projects

Selected Work

Distributed Systems

ElectroHub

An e-commerce marketplace built to scale. Consistent hashing for database sharding, Kafka for event streaming, Redis for rate limiting and real-time chat, RabbitMQ for async notifications, all running on Kubernetes.

PythonFastAPIKafkaRedisPostgreSQLKubernetes

Deep Learning · Healthcare

FetalVisionAI

0.000

AUROC · 77% sensitivity

Screens for congenital heart defects using ultrasound images and clinical data together. Cross-attention fusion, Focal Loss, Platt scaling. Deployed with drift monitoring and auto-retraining.

PyTorchGogRPCAirflowMLflow

Geospatial · Risk

Secure Routes — FIFA 2026

A route safety dashboard for the 2026 World Cup. Pulls from crime, border crossing, and migration datasets. Shows risk-adjusted routes on an interactive map.

PythonPostGISGeoPandasFlaskLeaflet.js

Explainable AI

Interview Performance Predictor

0.0×

Better accuracy than GPT-2

Predicts interview performance using MLP and Random Forest. SHAP + LLM reasoning explains every prediction in plain language, not just a score.

PythonScikit-learnSHAPMLPRandom Forest