CollectWise — AI Agent Engineer

davidjoseph-co; New York (NY); On-site; USD 200,000 - 300,000; Full time

Job summary

CollectWise in Manhattan, NY is hiring an AI Agent Engineer to build and own its voice AI agent infrastructure end-to-end on LiveKit, from orchestration to prompting strategy, across millions of consumer interactions.

You will design scalable real-time architectures, optimize latency, develop testing frameworks, and align technical decisions with business goals like containment and liquidation. This on-site role demands strong Node.js, AWS, and SQL skills and a hands-on startup mindset.

Qualifications

• 2+ years in voice AI or related fields.
• 3+ years in platform or backend engineering.
• Strong experience with Node.js, AWS, and SQL.
• Proven ability to ship production systems in startups or fast-paced environments.

Responsibilities

• Build the voice AI infrastructure on LiveKit from the ground up.
• Design real-time architecture: orchestration, state, latency, prompting.
• Own end-to-end prompting strategy and edge-case handling.
• Develop testing frameworks: A/B tests, regression, KPI reporting.
• Expand AI prompting to cover more debt collection scenarios.
• Improve containment, conversion, liquidation, using performance data.
• Optimize ASR, STT, TTS, and turn-taking for quality.
• Migrate client implementations to latest systems preserving customizations.
• Collaborate with product and leadership to align goals.

Skills

Prompt engineering

Analytics & ROI

Voice AI domain

Node.js

Latency optimization

Tools

LiveKit

AWS

SQL

SIP/telephony

Job description

CollectWise - AI Agent Engineer

Type: Full-time | On-site | Manhattan, NYCompensation: $200,000-$300,000 + 0.25%-1% equityHiring count: 1Visa sponsorship: None availableReports to: Sean, Founder & CEO (no LinkedIn provided)

About CollectWise

CollectWise is a YC-backed, seed-stage startup using generative AI to automate consumer debt collection, a $35B market in the US alone. Its AI agents handle the full late-stage recovery process - asset research, pre-legal outreach, litigation, credit reporting, and enforcement actions like garnishments and liens - and are already outperforming human collectors 2x at a fraction of the cost, running 10M+ calls per month.

Founded: 2024 | Team size: ~5 (Seed, 1-10) | Total funding: Not disclosed (YC-backed)Industry: AI ToolsWebsite: collectwise.comOffice: Manhattan, NY

Why Candidates Should Join

• Own it end-to-end: Build the voice AI infrastructure in-house from scratch on LiveKit and own the full stack - orchestration, prompting strategy, evaluation.
• Real scale, real outcomes: Millions of consumer interactions per month; the work ties directly to liquidation, containment, and conversion metrics, not just engineering metrics.
• Fast traction: $2M annualized run rate in a few months with a team of five, targeting $10M ARR this year.

Intake Call Summary

An intake video is present on the Contrario role page, but no transcript or written intake summary is included in the source HTML. Not fabricated - capture from the intake video separately if needed.

The Role

CollectWise needs an AI Agent Engineer to build and own its voice AI agent infrastructure end-to-end on LiveKit - the orchestration layer, the prompting strategy across millions of consumer interactions, and continuous performance improvement through rigorous testing and iteration. Equally suited to a deep voice AI engineer or a strong, analytical, business-minded infrastructure engineer.

What You'll Be Doing

• Build CollectWise's voice AI infrastructure on LiveKit from the ground up
• Design and own the architecture of real-time voice agent systems: orchestration, state management, latency optimization, prompting, and reliability
• Own the end-to-end prompting strategy: conversation flows, tone, compliance-safe phrasing, guardrails, and edge-case handling
• Build and maintain testing and evaluation frameworks: A/B testing, regression testing, conversation QA, benchmarking, and KPI reporting
• Write, expand, and continuously improve AI prompting to cover more real-world debt collection scenarios
• Iterate on performance data, call outcomes, and customer feedback to improve containment, conversion, liquidation rates, and customer satisfaction
• Optimize ASR, STT, TTS, turn-taking, interruption handling, and state transitions to improve voice quality
• Build scalable systems and tooling to support millions of voice interactions
• Migrate client implementations from legacy voice AI infrastructure to the latest systems while preserving client-specific customizations
• Work closely with product, operations, and leadership to translate business goals into high-performing voice agent behavior

Tech stack: LiveKit, Node.js, AWS, SQL; ASR/STT/TTS, real-time audio pipelines, SIP/telephony

Requirements

• 2+ years of experience in voice AI, conversational AI, prompt engineering, speech systems, or related agentic workflows
• 3+ years of experience in platform, infrastructure, or backend engineering
• Strong experience with Node.js, AWS, and SQL
• Experience building and shipping production systems in a startup or similarly fast-paced environment
• Experience with voice orchestration platforms or agent frameworks such as LiveKit or equivalent
• Strong prompt writing instincts and ability to design clear, effective, and resilient conversational flows for real-world edge cases
• Strong analytical skills and comfort tying technical work directly to business outcomes and ROI

Green Flags

• Voice AI experience in production, even from a no-name company or non-prestigious background.
• Has built AI infrastructure or agentic systems at a startup or AI company, understands what production-grade means at real scale
• LiveKit experience or direct experience with voice orchestration frameworks
• Business-minded, can connect technical decisions to liquidation rates, containment rates, and conversion metrics, not just engineering metrics
• Strong prompt engineering instincts, has designed conversational flows that handle edge cases gracefully under compliance constraints
• Infrastructure depth in Node.js, AWS, and SQL in a startup-paced environment where they owned what they built
• Experience with telephony infrastructure such as SIP trunking is a strong additional signal

Red Flags

• Pure academic or research background with no production deployment experience
• No analytical instincts, builds technically but cannot connect the work to business outcomes and ROI
• English communication skills not at native or near-native level. This is a hard filter due to the prompt engineering and compliance-sensitive consumer communication component
• Not willing to be in-person in Manhattan 4 to 5 days per week
• No experience with real-time systems or low-latency infrastructure

Role Details

• Experience - 3-10 years
• Salary - $200,000-$300,000
• Equity - 0%-1% (Job Details); outreach template quotes 0.25%-1%
• On-site policy - In-person, Manhattan 4-5 days/week
• Visa sponsorship - None available
• Employment type - Full-time
• Location - Manhattan, NY

Screening Questions

No screening questions section in the source HTML. Not fabricated.

Interview Process

Stage 1 - Pending Approval - Candidates awaiting initial approval.Stage 2 - Initial Call w/ Founder & CEO - First call with Sean.Stage 3 - Take-home AssignmentStage 4 - Technical InterviewStage 5 - Offer ExtendedStage 6 - Hired - Candidate accepts and starts.