All work
01 · Case study
OCTAGNT
An agentic AI platform that helps recruiters go from a hiring need to a shortlist: AI-written job descriptions, an autonomous voice interview agent and CRM automation.
- Role
- Full Stack Engineer
- Timeline
- 2024 - 2026
- Stack
- React
- Node.js
- TypeScript
- MongoDB
- OpenAI
- Twilio
- GCP
- AWS
- Docker
The problem
Recruiters spent hours on repetitive groundwork: writing job descriptions from scratch, running first-round screening calls, and copying client details between tools before any real hiring work began.
Approach
- 01Integrated OpenAI APIs to generate structured, on-brand job descriptions from a few inputs, with recruiter review before publishing.
- 02Built the AI Voice Interview Agent as a standalone microservice on GCP so it could scale and deploy independently of the core platform.
- 03Wired Twilio telephony to Google Speech-to-Text, Text-to-Speech and xAI Grok Voice for natural, low-latency screening conversations.
- 04Automated lead creation and client workspace provisioning through a Zoho CRM integration.
- 05Dockerized backend services and supported AWS deployments using S3, Lambda and CI/CD workflows.
Architecture
Challenges
- Keeping voice round-trips fast enough to feel conversational across telephony, speech and LLM providers.
- Isolating the voice agent so failures or load spikes never affect the main recruiter experience.
- Making AI output predictable enough for recruiters to trust and edit, not rewrite.
Result
Job description writing went from a manual chore to a review step, cutting that effort by roughly 70%, while screening interviews could run without a recruiter on the call.
- less manual JD effort
- ~70%
- independent voice-agent microservice
- 1
- AI & telephony providers orchestrated
- 4