Sait I.

I build production software for sensor data, numerical computing, and data platforms. I can help turn a Python prototype into an efficient C++ implementation, recover unreliable IMU data, or take a data/AI service from an idea to production. SENSOR DATA & PERFORMANCE - At Alma Technologies, ported a Python navigation stack to a SIMD-optimized C++ SDK. In my benchmarks, calibration dropped from 10 minutes to 3 seconds; tunnel tracking accuracy improved 5x over a u-blox baseline. - For a wearables startup, recalibrated corrupted inertial-sensor data, increasing usable training samples by 300%. Containerized ML experiments to reduce setup time from hours to minutes. - Completed a GPS-to-AR pose-alignment project on Upwork with a 5/5 client review for sensor fusion work. PRODUCTION DATA & AI SYSTEMS - At Matia, built and scaled a Python data-quality anomaly service from zero to hundreds of thousands of active monitors across customer data pipelines. - Initiated and shipped a customer-facing MCP server with OAuth and per-tool permissions. Built an AI triage agent processing about 100 on-call tickets a day using Datadog, source code, and change history. - Improved MongoDB data access and storage efficiency, enabling a lower cluster tier during rapid customer growth. - At Datatailr, developed C++ platform functionality for deployment and Slurm/Docker job execution; extended a Python interface to a C++ tick-data engine; implemented and benchmarked numerical routines for performance and accuracy. I work across algorithm development, profiling, implementation, and production integration. My background includes applied mathematics and physics at MIPT, high-energy physics research, and leading a seven-person engineering team. Core tools: C++, Python, PyTorch, NumPy, SIMD, TypeScript, MongoDB, Docker, Linux, AWS, Slurm, MCP, and OAuth. Based in Haifa, Israel; available remotely with US working-hour overlap. Open to substantial engineering projects and long-term collaborations. Tell me what you need to measure, accelerate, or bring to production.