Gen AI Engineers (Claude code, Codex, Cursor)

Accellor•San Francisco, CA

About The Position

Accellor is seeking an implementation-driven Gen AI Engineer with strong problem-solving ability, expertise with Claude Code, Codex, and Cursor, solid Python engineering skills, and hands-on understanding of machine learning, deep learning, GPUs, and CUDA fundamentals. The role may suit an early-career candidate who is technically sharp, curious, and able to learn complex AI systems quickly. The engineer will build models, debug training issues, optimize GPU workloads, understand tensor operations, and collaborate with research, platform, and engineering teams. The position involves developing and evaluating AI/ML models and training pipelines, working with deep learning architectures, diagnosing model and GPU issues, profiling and optimizing workloads, and building reliable AI inference and evaluation pipelines. It also includes implementing and documenting research ideas and communicating technical findings, trade-offs, and limitations. Accellor is an AI-native services firm that delivers AI, data, enterprise application, and product engineering offerings for industries including healthcare, life sciences, telecom, retail, financial services, and technology.

Requirements

Strong programming experience in Python.
Strong skills with Claude Code, Codex, and Cursor.
Hands-on experience with TensorFlow, PyTorch, or JAX.
Good understanding of data structures, algorithms, numerical programming, and clean code practices.
Strong fundamentals in deep learning, backpropagation, gradient descent, optimizers, loss functions, regularization, CNNs, Transformers, attention mechanisms, embeddings, and model evaluation.
Experience training and debugging models on CUDA-enabled GPUs.
Basic understanding of GPU memory, CUDA kernels, threads, blocks, warps, memory bandwidth, kernel launch overhead, mixed precision training, and CUDA out-of-memory debugging.
Comfort with Linux, Git, Docker, shell scripting, and experiment management.
Ability to reason from first principles and solve ambiguous technical problems.
Ability to build models, debug training issues, optimize GPU workloads, and work with tensor operations.

Nice To Haves

CUDA C/C++ or Triton kernel development.
C++ exposure for performance-critical AI systems.
Distributed training using PyTorch DDP, TensorFlow Distributed Strategy, JAX, NCCL, MPI, or similar tools.
Experience with GPU profiling tools such as NVIDIA Nsight Systems, NVIDIA Nsight Compute, PyTorch Profiler, or TensorFlow Profiler.
Experience with inference optimization, quantization, TensorRT, ONNX, XLA, or model compilation.
Exposure to LLMs, RAG systems, agents, embeddings, multimodal AI, or AI evaluation frameworks.
Understanding of FP32, FP16, BF16, INT8, tensor cores, and numerical stability.
Hands-on projects in model implementation, GPU acceleration, CUDA/Triton operators, distributed training, LLM evaluation, inference optimization, or research paper reproduction.
Ability to implement a model from a research paper, diagnose why a model is not learning, explain low GPU utilization, identify compute-, memory-, or communication-bound workloads, and write clean, reproducible, testable Python code.

Responsibilities

Design, build, train, and evaluate AI/ML models using Python and TensorFlow, PyTorch, or JAX.
Develop clean training pipelines with data loading, checkpointing, logging, validation, and experiment tracking.
Work on deep learning models including CNNs, Transformers, LLMs, embeddings, and attention-based architectures.
Debug model issues such as poor convergence, overfitting, unstable loss, NaNs, tensor-shape errors, and GPU memory failures.
Run and optimize models on GPUs, considering CUDA execution, memory usage, batching, mixed precision, and kernel performance.
Profile training and inference workloads to identify compute, memory, data-loading, and communication bottlenecks.
Support development of custom operators, fused kernels, CUDA/Triton-based optimizations, or framework-level performance improvements.
Build AI inference pipelines focused on latency, throughput, reliability, cost, and quality.
Create evaluation pipelines for model accuracy, robustness, hallucination risk, safety, latency, and regression testing.
Read research papers, implement ideas, run experiments, and document findings, trade-offs, and limitations.
Produce a large number of pull requests per month.
Work across model logic, GPU execution, data pipelines, and inference systems, and communicate technical findings clearly.