PR
Parandhama
tap to expand
Est. 2023Hyderabad · IN
Vol. INo. 001
// full-stack engineer, voice AI
hey, i'm

ParandhamaReddy

I build AI voice agents & the systems behind them

Full-stack engineer intern at Product Fusion, working on Tone, an open-source AI voice agent platform. I build retrieval pipelines, self-host speech and language models on GPU Kubernetes, and ship the Next.js front end around them. Based in Hyderabad.

Open for work · 2026

hold your cursor to it

The Profile — filed from Hyderabad

A bit about me

I got into programming the way most people do — I wanted to build something and Googled my way through it. That first project was terrible, but the feeling of making something work on screen was enough to keep going.

Today I'm a Full Stack Engineer intern at Product Fusion, working on Tone, an open-source platform for building AI voice agents. I built its retrieval pipeline on Docling and pgvector, selected embeddings with practical relevance, and kept the product focused on real agent workflows instead of demo-only UX.

I like the unglamorous middle of AI products: clean ingestion, honest latency numbers, autoscaling that lets a call survive a pod restart. On my own time I build things end to end to learn them properly, especially when they touch backend systems, infrastructure, or developer experience.

Studying Software Engineering at Kalvium, graduating 2027 with a CGPA of 9.4. Based in Hyderabad, India, and open to GenAI and full-stack roles.

“I like turning fuzzy ideas into things that actually ship.”

Currently

Building

Tone, an open-source AI voice agent platform, at Product Fusion

Studying

B.Tech Software Engineering, Kalvium (2023 to 2027) · CGPA: 9.4

Stack

Python, FastAPI, TypeScript, Next.js, Go, Kubernetes

Location

Hyderabad, India (UTC +05:30)

Available for work
The Record — career to date

Work and study

Current

Full Stack Engineer (Intern)

Product Fusion
May 2025 - Present
Hyderabad, Telangana
  • Working on Tone, an open-source AI voice agent platform built on FastAPI, Next.js, Pipecat and Kubernetes
  • Built the RAG pipeline: Docling parsing, OpenAI and Gemini embeddings, and pgvector retrieval, run as a background worker
  • Self-hosted STT, TTS and LLM models on GPU Kubernetes with vLLM and NVIDIA Riva, with KEDA autoscaling and drain-safe shutdown so live calls survive pod termination
  • Built the eval and benchmark layer: a DeepEval tool-call harness, latency benchmarks with percentiles, and per-call cost tracking
  • Added web calling over WebRTC and phone calling through Twilio and Telnyx, plus Salesforce and HubSpot as agent tools
  • Led a production hotfix in 2026, from root cause through deployment
PythonFastAPINext.jsPipecatpgvectorvLLMKubernetesKEDA
Current

B.Tech, Software Engineering

The Apollo University
2023 - 2027
Chittoor, Andhra Pradesh
  • CGPA 8.6 out of 10
  • Solo-built full-stack projects alongside coursework, including FarmCon, ClaimGuard and a Kubernetes-as-a-service API in Go
  • Trained and shipped SpamGuard, a spam classifier served through a Flask API
JavaPythonGoTypeScriptPostgreSQL
The Almanac — tools of the trade

My toolkit

Languages
Python
TypeScript
JavaScript
Go
Java
GenAI and voice
LLM APIs
RAG with Docling and pgvector
Agentic tool calling
Pipecat
STT and TTS pipelines
DeepEval
vLLM
Backend
FastAPI
Node.js
Express.js
Spring Boot
Frontend
React
Next.js
Tailwind CSS
Data
PostgreSQL
MongoDB
Redis
Prisma
Infrastructure
Docker
Kubernetes
GitHub Actions
Prometheus
Grafana
KEDA
Pulumi
By the Numbers — the commit ledger

Code & Contributions

0
Public Repos
0
Contributions
0
Stars Earned
0
Followers
0
Following
0
Public Gists

GitHub Achievements

Pull Shark×3
Quickdraw
YOLO
View Full Profile

Contribution Graph

Loading contributions...

Correspondence — write to the desk

Let’s talk

I'm open to GenAI and full-stack roles, interesting conversations, or just a friendly hello.

Drop a line below — I usually reply within a day.

rparandhama63@gmail.com