Bengaluru · Backend & Applied AI

Prateek Deshmukh

I build backend and AI services that ship to production: messy migrations, high-throughput pipelines, AI agents that talk to real systems, and developer-friendly APIs.

What I’m reading

Designing Data-Intensive Applications

Selected Work

Featured Projects

Hermes — LangChain Agent

Ingests WhatsApp webhooks, infers intent, and calls tools that update the internal ticketing system (status, ETA, assignee, priority) without human hand-holding.

LangChain LLM Automation

Backademia

MCP server exposing university portal data (attendance, marks, timetable) as tools. Includes auth, rate limits, and a REST layer for stable scraping sessions.

Python FastAPI SQLite MCP

Faff Chat

Real-time chat app with Socket.io plus semantic search using HuggingFace inference + SentenceTransformers for thread recall.

Socket.io Node.js HuggingFace NLP

Hooked

CLI wrapper around Ngrok for webhook debugging with header/key filters to isolate noisy payloads; cut integration debug time by ~50% during a payment migration.

Python Ngrok API CLI

Yaaru

Open-source, real-time take on “impostor who?” with a minimal high-contrast UI and hand-written WebSocket connection manager.

TypeScript React FastAPI WebSockets

prats.blog

Full-stack CMS with custom rich-text editor, live preview, JWT auth, and tuned uploads. Built to ship posts fast without cutting security corners.

React FastAPI SQLite JWT

Vapi Voice Outreach

Voice agents wired into Google Maps data to automate vendor outreach; optimized prompts/transcribers for regional accuracy.

Vapi Voice AI Automation
Professional Journey

Experience

Selected intern at high-growth teams and enterprise giants.

Engineering Intern

Sep 2025 — Mar 2026
Faff

Built recurring task service (cron-driven) for thousands of dependent tasks, LLM tools for the ticketing system, and ETA governance with deadline freezes/alerts.

Python FastAPI PostgreSQL LangChain

Incoming Software Engineer Intern

Summer 2026 (Expected)
NatWest

Selected as 1 of 11 interns from over 10,000 applicants (Top 0.1%) through rigorous technical evaluation.

Enterprise SWE FinTech

Full-Stack SDE Intern

May 2025 — July 2025
Easylib

Shipped documentation and observability for 300+ client instances; added modules across sales desk and cataloging following MVC patterns.

PHP CodeIgniter MS SQL MVC
The Toolkit

Tech I work with

Backend
PythonFastAPITypeScript (Hono)SQL
AI/Agents
LangChainLangGraphLangfuseVector DBs
Frontend
ReactNext.jsSvelteTailwind
Infrastructure
AWSCloudflare (Workers/R2/D1/KV)PostgreSQLRedisDockerGCP

SRM Institute of Science and Technology

B.Tech, Artificial Intelligence · 2023–2027 · CGPA 8.98/10

Systems & AI Focus

Core CS

Data Structures & Algorithms, Operating Systems, Networks

Data

DBMS, SQL, distributed data concepts

Applied Work

Building production agents and backend services alongside coursework.

Building dependable systems.

If you need backend or agent help, I’ll reply fast.