Senior AI Engineer

About the job

About AllysAI AllysAI is an AI Lab-as-a-Service company based in Dubai. We help enterprises implement AI solutions in 60–90 days — not 12+ months. Our clients include Al Futtaim, Merz Pharmaceuticals, and Abu Dhabi Government. We build production AI systems that enterprises actually use — not demos that sit on a shelf. About the Role You’ll be a core builder of AllysAI’s AI systems — both the platform (AllysOS) and the enterprise solutions we deliver to clients. This isn’t a research role. You’ll ship production AI that handles real enterprise workloads, solves edge cases that break other systems, and scales to meet the demands of government and Fortune 500 clients. We’re working on hard problems : multi-agent consensus, self-healing AI systems, complex multi-step reasoning over large knowledge bases, and fine-tuning models for domain-specific enterprise use cases. You’ll work directly with the founder and engineering team to push the boundaries of what’s possible in production AI. If you’re the type of engineer who reads papers on the weekend, has opinions on chunking strategies, and ships code that actually works at scale — this role is for you. What You’ll Do AllysOS Platform Development • Build and extend AllysOS — our proprietary AI platform that powers all client implementations • Design and implement new platform features : agent orchestration, memory systems, tool-use frameworks, and retrieval modules • Architect the multi-agent layer — build systems where multiple AI agents collaborate, reach consensus, and handle disagreements gracefully • Develop self-healing agent capabilities — agents that detect their own failures, recover from errors, and re-route when a step breaks • Build robust fallback mechanisms, retry logic, and graceful degradation for production reliability • Maintain and optimize existing AllysOS modules for performance, cost efficiency, and scalability RAG & Retrieval Systems • Design and build production RAG pipelines that handle enterprise-scale knowledge bases (millions of documents) • Implement advanced retrieval strategies : hybrid search (dense + sparse), re-ranking, query decomposition, and multi-hop reasoning • Build multi-step reasoning systems that chain retrieval, analysis, and generation across complex queries • Optimize retrieval accuracy through embedding model selection, chunking strategy refinement, and metadata filtering • Implement evaluation frameworks to measure and continuously improve RAG system quality (recall, precision, faithfulness, relevance) • Handle edge cases : ambiguous queries, conflicting information across sources, queries that span multiple documents, and out-of-scope detection LLM Fine-Tuning & Prompt Engineering • Fine-tune open-source and commercial LLMs for domain-specific enterprise use cases (pharma, government, automotive, retail) • Design and optimize prompt templates, system prompts, and instruction sets for consistent, high-quality outputs • Build prompt evaluation pipelines — automated testing of prompt variations across diverse inputs • Implement guardrails : output validation, toxicity filtering, hallucination detection, and compliance checks • Evaluate and benchmark models (GPT-4, Claude, Llama, Mistral, Gemini) for specific client requirements — cost, latency, accuracy trade-offs • Build model routing logic — directing queries to the right model based on complexity, cost, and quality requirements Agentic AI & Multi-Agent Systems • Build agentic workflows where AI systems plan, execute, and verify multi-step tasks autonomously • Implement agent consensus mechanisms — multiple agents evaluating the same problem and reaching agreement before acting • Design tool-use architectures : agents that call APIs, query databases, trigger workflows, and interact with external systems • Build agent memory systems — short-term context, long-term knowledge, and conversation state management • Implement human-in-the-loop checkpoints for high-stakes decisions and approval workflows • Design and test failure recovery : what happens when an agent gets stuck, hallucinates mid-chain, or hits an API timeout Enterprise Client Solutions • Build custom AI solutions for enterprise clients on top of AllysOS — chatbots, document processing systems, decision support tools, and automation workflows • Translate client requirements into technical architecture and deliver within 60–90 day timelines • Deploy and monitor AI systems in production — ensuring reliability, latency, and cost targets are met • Collaborate with the AI Data Engineer on knowledge base preparation and integration • Join client technical discussions to scope solutions, present architecture, and address concerns What We’re Looking For Experience • 4+ years in software engineering with at least 2 years focused on AI / ML or LLM-based systems • Production experience building RAG pipelines — not just prototypes, systems that handle real enterprise traffic • Hands-on experience with LLM APIs (OpenAI, Anthropic, Google) and open-source models (Llama, Mistral) • Built and deployed multi-agent or agentic AI systems in production • Experience fine-tuning language models for domain-specific tasks • Strong Python engineering — clean, tested, production-quality code with proper error handling • Experience with vector databases (Pinecone, Weaviate, Qdrant, ChromaDB) and embedding models • Cloud deployment experience (AWS, GCP, or Azure) — containerization, serverless, and API deployment Technical Skills • Deep understanding of transformer architectures and how LLMs work under the hood • RAG system design — you can architect a retrieval system from scratch and explain every design decision • Agent frameworks — LangChain, LlamaIndex, CrewAI, AutoGen, or custom agent architectures • API development — FastAPI, Flask, or similar for serving AI systems • Database proficiency — SQL, NoSQL, and vector databases • CI / CD and MLOps basics — version control, testing, deployment pipelines for AI systems • Monitoring and observability — tracking LLM costs, latency, error rates, and quality metrics in production Mindset • Builder mentality — you ship working systems, not slide decks about systems • Production-first thinking — you build for reliability, scale, and cost from day one, not as an afterthought • Ownership — you own your systems end-to-end, from architecture through deployment and monitoring • Curiosity — you stay current on the latest models, techniques, and frameworks and bring ideas to the team • Speed — you can prototype fast, validate quickly, and ship iteratively without over-engineering • Communication — you can explain complex AI concepts to non-technical stakeholders and clients Bonus Points • Experience building AI products for enterprise or government clients • Published research or open-source contributions in NLP, retrieval, or agent systems • Experience with Arabic NLP and multilingual AI systems • Knowledge of AI safety, alignment, and responsible AI deployment practices • Experience with real-time AI systems (streaming, WebSocket-based interactions) • Background in a specific vertical AllysAI serves : pharma, government, automotive, or retail • Contributions to or deep familiarity with major agent frameworks (LangChain, LlamaIndex, CrewAI) • Experience with evaluation frameworks : RAGAS, DeepEval, or custom LLM evaluation pipelines What We Offer • Competitive salary based on experience • Performance-based bonuses • Flexible remote work • Direct collaboration with the founder — you’ll shape the technical direction of AllysOS • Work on genuinely hard AI problems — multi-agent consensus, self-healing systems, enterprise-grade RAG • Build production AI for major enterprises — your code runs in systems used by thousands • Exposure to diverse industries and use cases — no two client projects are the same • Fast-paced, low-bureaucracy environment where your ideas ship quickly

Requirements

• RAG pipelines
• LLMs
• Python
• Agent frameworks
• API development
• Database proficiency
• MLOps

Preferred Technologies

• RAG pipelines
• LLMs
• Python
• Agent frameworks
• API development
• Database proficiency
• MLOps

Benefits

• Competitive salary based on experience
• Performance-based bonuses
• Flexible remote work
• Direct collaboration with the founder
• Work on genuinely hard AI problems
• Exposure to diverse industries and use cases

About the company

AllysAI is an AI Lab-as-a-Service company based in Dubai. We help enterprises implement AI solutions in 60–90 days — not 12+ months. Our clients include Al Futtaim, Merz Pharmaceuticals, and Abu Dhabi Government. We build production AI systems that enterprises actually use — not demos that sit on a shelf.