LLM Engineer

Vyro is at the forefront of innovation, transforming content creation through advanced AI and Machine Learning technologies. As a rapidly growing Gen-AI and SaaS-focused company, we empower creativity across industries with state-of-the-art tools. Our flagship products include ImagineArt, an AI-powered design studio that turns text into stunning visuals, and Chatly, an intelligent multi-modal assistant leveraging frontier AI models for seamless task management and idea generation.

With 15+ products, over 2.5 billion images processed, and 800,000+ daily active users, Vyro is actively shaping the future of creative tools. Join our passionate team of Vyronauts to make an impact and innovate with us!

Role Description

This is a full-time, on-site role for an LLM Engineer based in Islamabad. The role involves designing, developing, and fine-tuning LLMs, building agentic AI workloads, implementing data-driven algorithms, and deploying scalable solutions. You will collaborate closely with cross-functional teams to integrate cutting-edge machine learning capabilities into Vyro’s products, while exploring new methods to enhance performance, reliability, and efficiency.

Qualifications

• 4+ years of industry experience in Machine Learning or NLP
• Bachelor’s degree in Computer Science (BSCS) or a related field

• Deep experience leveraging closed-source SOTA models from OpenAI, Anthropic, Google, and xAI
• Strong understanding of complex reasoning, tool-use, and multi-step AI pipelines

• Expert grasp of transformer variants and Mixture-of-Experts (MoE) architectures
• Proven hands-on experience with open-weight SOTA models such as Llama 3.x, Mistral Large, Qwen 2.5, Phi-4, etc.

• Mastery of multi-agent orchestration using frameworks like LangGraph (stateful agents), AutoGen, or CrewAI
• Experience implementing DSPy for declarative, self-optimizing prompt pipelines

• Implementation experience with GraphRAG and hybrid retrieval strategies
• Expertise with vector stores (Qdrant, Milvus, Weaviate) and semantic caching for long-term agent memory

• Experience deploying high-throughput models using vLLM, TensorRT-LLM, or SGLang
• Familiarity with FlashAttention-2, KV caching, and quantization techniques (AWQ, EXL2)
   

• Work on innovative AI products like Chatly and ImagineArt that are shaping the future of user interaction and creativity
• Collaborate with a passionate, talented team that values experimentation, innovation, and data-driven decision-making
• Competitive salary and benefits package
• A growth-driven culture that encourages learning, ownership, and continuous improvement

Note: This is an onsite position at our office in H12, Islamabad, for residents of Pakistan. Candidates residing outside of Pakistan may be considered for remote work opportunities.