Software Engineer · Voice AI · Open Source

Geoffrey
Robinson

I build reliable realtime systems and product experiences. Most recently, I worked on production voice agents, telephony infrastructure, and AI integrations at Husky Voice AI.

GR

Python · TypeScript · LiveKit · AWS · Docker · C++ · Distributed systems

01 / Experience

Building voice AI in production

Husky Voice AI

Software Engineering Intern

March 2026 — August 2026 Bengaluru

Python
TypeScript
LiveKit
Docker
AWS
MongoDB
  • Worked directly with the CTO in a fast-paced startup team, using Agile/Scrum practices and Jira to plan and deliver product work.
  • Built and deployed production LiveKit workers for concurrent inbound and outbound voice calls, with lifecycle management, failure recovery, and horizontal scaling.
  • Implemented secure BYOK infrastructure that routes LLM, speech-to-text, and text-to-speech requests to each tenant’s configured providers.
  • Engineered agent-to-human transfers across LiveKit, Asterisk, and PSTN with retry, timeout, emergency escalation, and call-context preservation.
  • Integrated Twilio and Interakt for WhatsApp plus Telnyx for SIP and voice through a shared provider abstraction.
  • Built crash-resilient per-call diagnostics with incremental S3 multipart streaming, secret redaction, and a zero-loss JSONL fallback.
  • Shipped natural-language callback scheduling with notice windows and duplicate-call protection directly into the existing campaign scheduler.

02 / Selected work

Systems built to be useful

Featured project

C++ Load Balancer

01

A concurrent HTTP load balancer with pluggable backend-selection strategies, persistent health tracking, and resilient request forwarding.

  • Thread-safe round-robin selection with a bounded worker queue.
  • SQLite-backed registration and health state with automated checks.
  • Backend health filtering and HTTP request forwarding.
C++17/20
SQLite
Multithreading
HTTP

03 / Open source contributions

Code contributed in the open

AI memory · 2 merged PRs

Cognee

01
  • Created a schema-only relational ingestion pipeline that builds a knowledge graph for Text-to-SQL tasks while reducing required data ingestion.
  • Connected Tavily web search retrieval to Cognee’s memory system.
Python
Docker
View my PRs

Data infrastructure · 2 merged PRs

CocoIndex

02
  • Added custom PostgreSQL schema support for CocoIndex’s internal tables, improving isolation in shared databases.
  • Strengthened target-state key typing by requiring keys to conform to stable key types.
Python
Rust
PostgreSQL
View my PRs

04 / Toolkit

Across the full stack

Languages

Python
TypeScript
C#
JavaScript
Java
C
Modern C++17/20

Frontend

Vue 3
Vite
Vue Router
Tailwind CSS
React
Next.js

Backend & Data

Node.js
Express.js
FastAPI
.NET
Django REST Framework
Flask
REST APIs
PostgreSQL
MongoDB
SQLite
Redis

Voice & Realtime

Asterisk
SIP/PJSIP
WebRTC
LiveKit
Twilio
Telnyx

Cloud & Engineering

AWS (EC2, ECR, App Runner, S3, RDS, Lambda)
Docker
Linux
Distributed Systems
Socket Programming
Multithreading
TDD
Agile/Scrum
Jira
Code Reviews

05 / Education

The foundation

RV Institute of Technology and Management

Bachelor of Engineering · Information Science and Engineering

September 2022 — September 2026 (expected)Bengaluru, Karnataka

06 / Recognition

Recent wins

Winner — Contribute To Win 2025

Cognee · Fall 2025

Star Performer

Husky Voice AI · May 2026

Let’s build something

Start a conversation

I’m interested in software engineering, realtime systems, voice AI, and open-source opportunities.