AI/ML Software Engineer

About

Humai exists because we believe AI should solve real problems, not create new ones. We build products for the specific moments where AI actually makes life easier — whether that's helping a business streamline how they work or helping someone make better decisions in their personal life. Enterprise and consumer, same principle: if it doesn't make a tangible difference, we don't ship it. Small team in Dubai, real products in market, growing fast.

Open Positions

1) AI Engineer – Voice, Matching & Multimodal

You handle the hardest signals. Turning messy audio into data, building algorithms that match people, and working across modalities.

Skills you’ll use:

• Voice & Speech AI – Real-time speech-to-text pipelines, speaker diarization, multilingual ASR, and streaming audio processing for business conversations
• Recommendation & Matching Systems – Matching and ranking algorithms for two‑sided marketplaces. Real-time inference, feature engineering, feedback loops, and fairness constraints
• Multimodal AI – Systems that combine text, audio, image, and video for richer understanding — the direction frontier AI is moving and where our products are headed

2) AI Applications Engineer – Agents & Language

You build the intelligence users interact with. Agents that take action, systems that understand language, features that ship.

Skills you’ll use:

• AI Agents & Workflows – Design autonomous agents that use tools, make decisions, and complete multi‑step workflows across products — CRM automation, proactive coaching, recruiting pipelines
• LLM Applications & Traditional ML – Build intelligent features using LangChain, agent frameworks, and classical ML (classification, ranking, clustering) where they outperform LLMs
• NLP & Conversational AI – Intent classification, entity extraction, summarization, and dialogue systems. Multilingual capability (English + Arabic) is a strong plus

3) Senior AI/ML Engineer – Models & Infrastructure

You own the AI backbone. From training models to running them reliably in production.

Skills you’ll use:

• LLM Systems & Architecture – Design and optimize LLM‑powered systems, RAG architectures, and orchestration pipelines (DSPy, LangGraph)
• Model Fine‑Tuning & Training – Fine‑tune foundation models (Llama, Mistral, Whisper, etc.) for domain‑specific tasks. Build training pipelines, curate datasets, and run evaluations
• MLOps & ML Platform – Own model serving, monitoring, experiment tracking, feature stores, and CI/CD for ML. Make sure models run reliably at scale, not just in notebooks

What’s true across all roles

• Stack: Python, GCP, PostgreSQL (pgvector), Redis, Docker, Cloud Run, GitHub Actions
• Location: Full‑time, on‑site in Dubai (DMCC Free Zone)
• Team: You work directly with the founding team — no layers
• Pace: Ship weekly. Production over prototypes.

What we look for

• You’ve shipped ML/AI to production — not just notebooks
• You take ownership and don’t wait to be told what to build
• You’re honest about what works and what doesn’t
• You’re excited about building products, not just models