AI Engineer - Agentic AI & Voice AI and Data Science

AI Engineer - Agentic AI & Voice AI and Data Science

Company

Dasceq

Location

Noida / Gurugram preferred (Hybrid – Delhi NCR)

Experience

3-8 Years

Employment

Full-time

Proposed Salary

₹ 40,000–₹ 1,00,000 per month

About Dasceq

Dasceq is building a next-generation Agentic AI platform that combines AI-powered conversations, predictive analytics, intelligent workflow automation and business-system integrations.

The platform runs inbound and outbound AI agents across voice calls, SMS and email, with automated dialing, CRM integrations, payment workflows and campaign management. We serve accounts receivable and debt collection businesses and are expanding into home services, field dispatch, sales, marketing and customer support automation.

Role summary

We are looking for a hands-on AI Engineer with strong Python programming skills and practical experience in building AI/ML applications, LLM-powered systems or conversational AI solutions.

The candidate should understand how Agentic AI applications work internally and should be capable of developing their components, integrating APIs, writing backend code, debugging workflows and contributing to the development of an AI platform from the ground up.

We are particularly interested in candidates who have actually built AI agents, voice bots, conversational AI applications or automation systems rather than candidates who have only used existing AI platforms or no-code tools.

Key responsibilities

• Architect and build agentic voice platforms from scratch

Design and code the end-to-end pipeline for inbound and outbound AI calls: telephony, speech-to-text, LLM reasoning, text-to-speech and call state management.

Build agent orchestration with tool calling, multi-step workflows, conversation memory, guardrails, fallbacks and human handoff.

Optimise real-time performance: latency, barge-in and interruption handling, silence and turn detection, streaming.

Collections and inbound call workflows

Implement collections flows: identity verification, disclosures, negotiation, promise-to-pay, payment links, objection handling, dispute and callback handling.

Build inbound flows for payments, account queries, scheduling and routing.

Build in compliance controls such as call-time rules, consent, opt-out, DNC handling, call recording and audit logs (FDCPA, TCPA, Reg F awareness for the US market).

Telephony, APIs and integrations

Integrate SIP, VoIP and CPaaS providers (Twilio, Telnyx, Vonage, Plivo, Asterisk or FreeSWITCH or equivalent) and dialers, with SMS and email channels.

Build REST APIs and webhooks that connect agents to CRM, payment gateways, scheduling and collections systems.

Build robust retry, idempotency and error-handling logic for third-party integrations.

LLM engineering and evaluation

Engineer prompts, structured outputs and RAG pipelines for domain-specific conversations.

Build evaluation frameworks

conversation quality scoring, hallucination and tool-failure testing, regression suites and call-transcript analysis.

Select, benchmark and swap LLM, STT and TTS vendors on cost, latency and accuracy.

Data science and predictive analytics

Build and monitor models for propensity-to-pay, contactability, best time to call, recovery and segmentation.

Perform data cleaning, feature engineering and model evaluation on payment and interaction data; support A/B tests and campaign analytics.

Cloud, deployment and production ownership

Deploy and run services on AWS (EC2, Lambda, S3, RDS, CloudWatch) using Docker and CI/CD.

Own logging, monitoring, call-failure debugging and performance tuning in production.

Follow credential management, PII protection and cloud-security practices.

Engineering practice and client delivery

Write clean, tested, documented Python code; use Git; review code and take part in design discussions.

Join client requirement sessions with senior engineers; support UAT, demos and onboarding; resolve client-reported issues.

Must-have skills and experience

2–5 years of software or AI engineering experience, with strong Python (async programming, FastAPI or Flask, WebSockets, testing).

Has built at least one agentic or conversational AI system from scratch, voice preferred, that handled real calls in production. Candidates who only configured no-code or low-code tools (for example Vapi or Retell setups without owning the underlying code) will not match.

Hands-on LLM engineering: OpenAI, Anthropic or Gemini APIs, tool/function calling, prompt design, structured outputs, conversation memory and agent frameworks (LangChain, LangGraph, or custom orchestration).

Real-time voice stack experience: STT and TTS providers (Deepgram, AssemblyAI, Whisper, ElevenLabs, Cartesia or similar), streaming audio, latency optimisation.

Telephony integration experience: SIP, Twilio or equivalent, webhooks, call flows, dialers, IVR or call routing.

Experience with inbound or outbound call-centre, collections, lending, BPO or customer-support automation use cases.

Strong SQL and data skills

Pandas, NumPy, scikit-learn; ability to build and evaluate predictive models.

AWS (EC2, Lambda, S3, RDS, CloudWatch), Docker, Git and CI/CD in a production setting.

Clear written and spoken English; able to explain design decisions and document systems.

Good to have

Open-source voice frameworks such as Pipecat, LiveKit Agents or Vocode; Asterisk or FreeSWITCH.

RAG, vector databases, fine-tuning or running open-source models.

Knowledge of collections regulations and compliance (FDCPA, TCPA, Reg F) or lending operations.

CRM and payment gateway integrations (Salesforce, HubSpot, Stripe, Razorpay, Authorize.net or similar).

Redis, Postgres, message queues, Kubernetes, observability tools.

A public GitHub repo, demo video or write-up of an agent you built.

Education and ideal candidate

BE/BTech, BSc/MSc, MCA or equivalent in Computer Science, Data Science, AI, Mathematics, Statistics or a related field. Strong shipped work outweighs the degree.

The ideal candi