AI Software Engineer

a. Job Description: We are looking for an AI Software Engineer with expertise in performance profiling, model fine-tuning, and architecture identification for large-scale AI models such as Llama, DeepSeek, VLM, Evo-2, LLMs, Robot VLM, and DNA-based models. The ideal candidate has experience with deep learning frameworks, hardware acceleration, and AI optimization techniques to enhance the efficiency and scalability of AI models. b. Performance Profiling & Optimization Analyze and optimize AI model performance across various hardware platforms (GPUs, TPUs, NPUs). Profile training and inference pipelines for memory usage, compute efficiency, and latency. Work with CUDA, TensorRT, PyTorch, and JAX to optimize models for production deployment. Implement quantization, pruning, distillation, and mixed-precision training techniques. Debug performance bottlenecks using NVProf, Nsight, TensorBoard, and PyTorch Profiler. 2. Architecture Identification & Model Analysis Reverse-engineer and analyze LLM/VLM architectures to extract key architectural details. Identify activation functions, attention mechanisms, and parameter distributions in pretrained models. Compare performance trade-offs between transformer-based, mixture-of-experts (MoE), and diffusion models. Work on model compression techniques for real-time AI applications. 3. Fine-Tuning & Customization Fine-tune LLMs, VLMs, and DNA-based AI models on domain-specific datasets. Use LoRA, QLoRA, PEFT, and Adapter methods for parameter-efficient fine-tuning. Implement prompt engineering, retrieval-augmented generation (RAG), and reinforcement learning (RLHF) for LLMs. Train and deploy robotic VLMs for real-world AI applications. c. Qualifications: Bachelor’s/Master’s/PhD in Computer Science, Machine Learning, AI, or related fields. 3+ years of experience in deep learning, model optimization, and AI performance engineering. Strong proficiency in Python, PyTorch, TensorFlow, and JAX. Experience with CUDA, Triton, TensorRT, and ONNX for AI acceleration. Understanding of transformer architectures, vision-language models (VLMs), and multi-modal AI. Familiarity with LLM fine-tuning techniques, large-scale distributed training, and RLHF. Experience working with scientific AI applications (e.g., DNA sequencing, robotics, computational biology). - Non smoking