May 2023 – Jul 2023 · 2 mos · Bengaluru, Karnataka, India · Hybrid

• Developed an internal accuracy analysis framework for quantized ONNX models, enabling engineers to evaluate layer-wise performance degradation after quantization
• Built APIs for per-layer error analysis using metrics like MSE, helping identify accuracy bottlenecks and improve model optimization workflows
• Improved the efficiency of model evaluation pipelines, enabling faster experimentation and debugging for deep learning model deployment