Ayush Kaushik · Backend & AI Engineer
Backend systems that stay up. AI systems that ship.
I build and run production APIs, RAG pipelines, and the data infrastructure behind them. Mostly Python and FastAPI. I care about the boring things: reliability, performance, and code the next engineer can read.
- FastAPI
- production specialist
- AI / RAG
- LLM systems
- 4+ yrs
- running production systems
What I do
Engineering across the full backend stack
Four areas where I do my best work. The interesting projects usually need all of them.
Backend Development
Python and FastAPI services with clean architecture, async done right, and the reliability work that keeps them running at 3am.
- FastAPI & async Python
- SQLAlchemy & Postgres
- Production hardening
API Architecture
Well-designed REST APIs that scale: versioning, auth, rate limiting, observability, and contracts your consumers can rely on.
- REST & OpenAPI design
- Auth & rate limiting
- Caching & scaling
AI / LLM Engineering
RAG pipelines and LLM features that are grounded and evaluated. Built to survive production, not just the demo.
- RAG & vector search
- LLM app integration
- Evaluation & guardrails
Data Pipelines
Ingestion, transformation, and scheduling that move data reliably, from one-off scripts to monitored production workflows.
- ETL & ingestion
- Scheduling & jobs
- Cloud (GCP) automation
Experience & projects
What I've built, and where
3+ years shipping data systems, now focused on backend & AI engineering in production.
Featured project
Medyx: Healthcare Platform Backend
Python Backend Developer · Scalability Engineers
Production backend powering a healthcare platform: FastAPI services, async jobs, and a notification system, deployed and scaled on Google Cloud.
- Production-grade REST APIs in FastAPI for core clinical workflows
- Async services, scheduled jobs & background tasks for reliable notifications
- SQLAlchemy + PostgreSQL/MySQL: schema design, migrations, query tuning
- JWT/OAuth role-based auth; Firebase/FCM push with retry & failure tolerance
- Deployed on GCP (Cloud Run, GKE, BigQuery) with CI/CD and observability
Timeline
-
Python Backend Developer
Mar 2025 – PresentScalability Engineers
Building scalable, secure backend services for the Medyx healthcare platform: APIs, async workflows, and cloud infrastructure.
-
Data Analyst / Data Engineer
May 2022 – Mar 2025Scalability Engineers
BI dashboards (Qlik Sense, Power BI), optimized T-SQL, end-to-end ETL pipelines, and Python/PowerShell automation.
-
Author
Jan 2026 – PresentLogicLoopTech
Writing in-depth backend & AI engineering articles for developers.
B.Tech, Information Technology, Uttaranchal University
Selected work
Deep dives into real production problems
No case-study theater. These are full technical breakdowns of issues I've actually debugged and shipped fixes for.
FastAPI SQLAlchemy Session Leak Detection: Diagnose and Fix Long-Running DB Sessions in Production
Diagnosing a production failure that only surfaces under sustained load, and the connection-pool fix that resolved it.
Read the breakdown 02FastAPI in Production: The Complete Deployment Guide (Docker, Workers, Scaling & Best Practices)
The full path from local development to a hardened, scalable production deployment with Docker and workers.
Read the breakdownFrom the blog
Latest engineering writing
AI agent python ollama: Build, Test, Deploy with FastAPI
Learn how to run an AI agent in Python with Ollama, from local installation to FastAPI integration, testing, and production scaling, using code examples.
AI Generated Code Detection: Practical Guide for Builders
Learn how to reliably detect AI-generated code in CI/CD, spot hidden security bugs, and navigate legal risks before they break production.
Building microservices python fastapi: design to production
Build microservices python fastapi from design to production, covering REST/gRPC choices, RabbitMQ/Kafka queues, Docker/K8s deployment, testing and security.
Working on something similar?
If you're building backend or AI systems and want a second set of senior eyes, let's talk.