AI Voice Agent Engineer – Reinforcement Learning

Full Description

Role Summary

We are building an advanced AI-powered cold-calling agent capable of natural, human-like conversations, personalized decision-making, and adaptive behaviour.

We already have a working agent.

Your role is to

train, optimize and evolve its intelligence

.

You will design the agent's

behaviour model

,

tone

,

response patterns

, and

learning logic

using RL, conversational AI techniques, and structured training datasets.

This position requires strong ownership and the ability to convert business rules into AI behaviour.

Key Responsibilities

Behaviour & Reinforcement Learning

• Build reward/penalty frameworks for call outcomes.

• Train the agent to improve over time (RLHF / RLAIF).

• Shape tone, pace, personality, and decision logic.

• Create policies for "how the agent should act" in each scenario.

2. Knowledge Source Management

Train the agent on

how to choose between

:

• Live website information

• Internal knowledge base (RAG)

• Custom LLM logic

• Pre-written scripts

• Dynamic call context

Define

which data should be used when

so the agent answers accurately.

Voice Agent Optimization

• Reduce voicemail ratio

• Improve pickup-to-conversation conversion

• Adjust speaking speed, pauses, natural flow

• Reduce hallucination or wrong info

• Improve call stability

Data & Evaluation

• Analyse call recordings

• Detect behavioural issues

• Recommend training datasets

• Build continuous learning cycles

Required Expertise

• Strong experience in

Reinforcement Learning (RLHF/RLAIF)

• Experience with

Vapi / Twilio / SIP telephony

(preferred)

• Strong knowledge of

LLM behaviour modeling

• Knowledge of

RAG pipelines

, embeddings & vector databases

• Experience building

voice agents

or conversational systems

• Ability to translate business rules into AI logic

• Comfortable working in fast-moving environments

Soft Skills

• Clear communication with non-technical leadership

• Ownership mindset

• Ability to document logic in simple terms

• Strong problem-solving orientation