Inference Optimization Engineer(LLM and Runtime)

Location :   Bengaluru, Karnataka

About the Company :

Sustainability Economics.ai is a global organization, pioneering the convergence of clean energy and AI, enabling profitable energy transitions while powering end-to-end AI infrastructure. By integrating AI-driven cloud solutions with sustainable energy, we create scalable, intelligent ecosystems that drive efficiency, innovation, and long-term impact across industries. Guided by exceptional leaders and visionaries with decades of expertise in finance, policy, technology, and innovation, we are committed to making long-term efforts to fulfil this vision through our technical innovation, client services, expertise, and capability expansion.

Role Summary :

We are seeking a highly skilled and innovative  Inference Optimization (LLM and Runtime)  to design, develop, and optimize cutting-edge AI systems that power intelligent, scalable, and agent-driven workflows. This role blends the frontier of generative AI research with robust engineering, requiring expertise in machine learning, deep learning, and large language models (LLMs) and latest trends going on in the industry. The ideal candidate will collaborate with cross-functional teams to build production-ready AI solutions that address real-world business challenges while keeping our platforms at the forefront of AI innovation.

Key Tasks and Accountability :

• Optimization and customization  of large-scale generative models (LLMs) for efficient inference and serving.
• Apply and evaluate advanced  model optimization techniques  such as quantization, pruning, distillation, tensor parallelism, caching strategies, etc., to enhance model efficiency, throughput, and inference performance.
• Implement  custom fine-tuning pipelines  using parameter-efficient methods (LoRA, QLoRA, adapters etc.) to achieve task-specific goals while minimizing compute overhead.
• Optimize  runtime performance  of inference stacks using frameworks like vLLM, TensorRT-LLM, DeepSpeed-Inference, and Hugging Face Accelerate.
• Design and implement  scalable model-serving architectures  on GPU clusters and cloud infrastructure (AWS, GCP, or Azure).
• Work closely with platform and infrastructure teams to reduce  latency, memory footprint, and cost-per-token  during production inference.
• Evaluate  hardware–software co-optimization strategies  across GPUs (NVIDIA A100 / H100), TPUs, or custom accelerators.
• Monitor and profile performance using tools such as  Nsight, PyTorch Profiler, and Triton Metrics  to drive continuous improvement.

Key Requirements :

Education & Experience

Skills

What You’ll Do

What you will bring

Why Join Us