Mobiloitte
Website:
mobiloitte.com
Job details:
Technical Lead – AI Solutions & GTM (Voice AI · RAG · Full Stack)
Location: Okhla Phase-I | Work from office
Experience: 7+ years
About Mobiloitte Group
Mobiloitte Group is a technology group headquartered in Delhi, India, with entities in the USA, UK, UAE, Singapore and South Africa. The group builds AI-first products and delivers enterprise engineering for clients across these markets. Its flagship product, Converiqo.AI, is a multi-tenant AI customer-engagement platform. It runs conversations, automation and outbound campaigns across WhatsApp, web chat, Instagram, Messenger, Telegram, SMS and voice, for businesses in India, the USA, the UAE and South Africa.
About the Role
This is the engineer our customers meet. You will be the technical face of Converiqo.AI in sales conversations:
- Running discovery with the prospect
- Building the demo on the prospect's own data
- Designing the deployment
- Committing to numbers you can defend
Then you will lead the team that delivers it.
It is a hands-on role. You will write production code, review your team's code, and debug live systems. We are not looking for a pre-sales speaker without build depth, nor for an engineer who cannot face a client.
Not a fit: experience limited to prompt engineering, wrapping third-party AI APIs, or configuring no-code voice/chatbot platforms without building the pipeline underneath.
What You Will Do
Go-to-market and solutioning
· Run technical discovery with prospects. Map their use cases to platform capabilities and call out gaps honestly.
· Build live demos and time-boxed proofs of concept on client data, such as knowledge-base ingestion, WhatsApp flows and voice agents, in days rather than weeks.
· Design solution architecture for client deployments:
o dedicated stack per customer
o cloud, VPC or on-premises hosting, and data residency
o integration with CRMs, e-commerce stores (WooCommerce, Shopify) and telephony
· Own the technical sections of proposals, RFPs, SOWs, security questionnaires and data-processing agreements. This includes sizing, latency and cost-per-conversation estimates, and SLAs.
· Bring field learnings and competitor teardowns back into the product roadmap.
Build and deliver
- Voice AI: build streaming STT → LLM/RAG → TTS pipelines with barge-in, endpointing, SIP/telephony and WebSockets. Implement outbound calling controls: consent, Do-Not-Call, calling windows and emergency stop.
- RAG: cover document parsing, chunking quality, embeddings, hybrid search, reranking, and grounding and citations. Make sure the system says "I don't know" when it should, and keep evaluation sets that catch regressions.
- Agents: build bounded agentic workflows using LangGraph, MCP or equivalent. Have the judgement to keep agent loops off latency-critical paths.
- Private LLMs: deploy Llama, Qwen, Mistral or similar on vLLM, TensorRT-LLM or Ollama. Handle GPU/VRAM sizing, quantization, concurrency, and routing between hosted and self-hosted models.
- Cost and performance: per-turn token and cost metering, latency budgets, caching, observability and alerting.
- Security: tenant isolation, PII encryption, role-based access, secrets management and webhook signature verification.
Lead
- Lead a 25+ engineering team: architecture decisions, code reviews, merging module-scoped work, and engineering standards.
- Use AI coding agents (Claude Code, Codex or similar) productively. Scope the work tightly, review the output, and verify against the code rather than the agent's report.
- Mentor engineers and leave written decisions behind.
Must Have
- 7+ years in software engineering, with at least 2 years shipping LLM features used by real customers in production.
- Personally built and run in production at least one of: a voice bot, an enterprise RAG system, or a self-hosted LLM deployment. Two or more is strongly preferred.
- Strong Python (FastAPI or similar) and PostgreSQL, plus working proficiency in TypeScript with React / Next.js.
- Docker, CI/CD, and at least one of AWS, Azure or GCP. Kubernetes is a plus.
- Client-facing experience in discovery, demos, PoCs or solution design, and the ability to explain trade-offs to both CXOs and engineers.
- Clear written English for proposals and architecture documents.
- Evidence-driven: you measure quality and latency with numbers, not impressions.
Good to Have
- WhatsApp Business Platform / Meta Cloud API; Instagram and Messenger APIs.
- Vector databases (Qdrant, pgvector, Milvus, Weaviate) and Redis.
- Telephony and real-time media: Twilio, Plivo, Exotel, Asterisk / FreeSWITCH, LiveKit or Pipecat.
- Hindi / Hinglish and other Indian-language STT and TTS.
- Compliance exposure: India's DPDP Act, TRAI telemarketing rules, GDPR, SOC 2.
- Replacing public LLM APIs with self-hosted models in production.
- E-commerce integrations and AI shopping assistants.
- Open-source contributions in the AI / voice space.
Click on Apply to know more.
