Job Description

We are hiring an experienced

AI/ML Architect

to lead the design, deployment, and optimization of an

on-premise, multilingual AI-powered call center platform

for a strategic UAE federal government entity. This project supports over

25,000 voice calls daily

and leverages

LLM, STT, and TTS technologies

to deliver responsive, high-availability citizen services across six languages.

As the AI/ML Architect, you'll take ownership of

end-to-end AI system design

, manage

LLM fine-tuning on large-scale datasets

, and collaborate cross-functionally to ensure sub-250ms response times under peak load--all within

UAE data residency and compliance

frameworks.

Key Responsibilities:

Architecture & Design:

Architect scalable pipelines for real-time

STT

,

TTS

, and

LLM

systems for high-concurrency telephony workflows (150+ concurrent calls). Design and implement optimized low-latency inference infrastructure for real-time multilingual voice interaction.

LLM Fine-Tuning & Deployment:

Fine-tune and deploy

open-source LLMs

(e.g., Falcon, LLaMA, Mixtral) on local compute using UAE-specific and Client's-provided datasets. Ensure secure, offline training (no cloud access) with full on-prem data privacy compliance.

Data & Knowledge Integration:

Design

RAG (Retrieval-Augmented Generation)

pipelines over structured and unstructured data (e.g., PDFs, websites, databases). Collaborate with data engineers to index, embed, and retrieve content from 1-10TB knowledge bases.

Multilingual Speech AI Integration:

Integrate multilingual

Speech-to-Text

(Whisper, Coqui) and

Text-to-Speech

(XTTS, Bark) models with optimized latency (<250ms STT, <200ms TTS). Customize voices using Arabic (Emirati), English, Urdu, Hindi, French, and Tagalog samples provided by ICP.

Optimization & Benchmarking:

Continuously optimize AI model inference, reduce memory/compute overhead, and maintain high availability. Design fallback, failover, and versioning strategies for STT/TTS/LLM models.

Cross-Functional Collaboration:

Work closely with DevOps, backend teams, compliance officers, linguists, and telecom engineers for end-to-end platform delivery. Support the QA process and post-deployment model updates and audits.

Required Qualifications:

Master's or PhD in

AI, Machine Learning, Computational Linguistics

, or related field. 6+ years of experience in

AI/ML architecture

, with 3+ years deploying

production-grade LLM/NLP/STT/TTS

systems.

Hands-on expertise with:

LLM fine-tuning (e.g., Falcon, LLaMA, Mistral) STT/TTS models (Whisper, XTTS, Bark, Coqui) RAG pipelines (FAISS, Milvus, Weaviate, LangChain/Haystack) Deep Learning frameworks (PyTorch, TensorFlow) Tokenizers, quantization (GGUF/GGML), distributed training Experience with large-scale dataset curation and vector embeddings. Deep understanding of

Arabic dialects (especially Emirati)

and multilingual NLP systems is a major plus.

Bonus Skills:

Experience with telephony systems and SIP-based call routing Familiarity with UAE data protection regulations and AI Ethics Charter Experience in government or public-sector deployments Knowledge of A/V pipelines, audio preprocessing, and latency profiling

What We Offer:

Work on one of the most ambitious

sovereign AI deployments in the UAE

High-impact leadership role in a

national-scale project

Opportunity to shape AI-powered public service delivery Long-term growth with SPAI's expanding government AI portfolio
Job Type: Full-time

Pay: AED30,000.00 - AED40,000.00 per month