Computer Vision & C++ Engineer (Non AI)

Job Description:

Key Skills: Computer Vision, Image Processing, C++

• 7–10+ years of experience in computer vision, image processing, or scientific/medical imaging software development.
• Proven experience delivering performance‑critical C++ algorithmic solutions.
• Strong expertise in classical computer vision and image processing (non‑ML).
• Advanced proficiency in C++ (C++14/17 or newer).
• Hands‑on experience with Intel IPP and/or Intel MKL.
• Proficiency in Python for prototyping and validation.
• Solid understanding of:

       Signal processing concepts

• Numerical optimization
• Performance profiling and optimization techniques

Domain Knowledge (Preferred)

• Experience with OCT, retinal imaging, or biomedical image processing.

• Familiarity with retinal anatomy (ILM, RNFL) and segmentation challenges.
• Meets or exceeds all segmentation accuracy requirements defined per retinal region.

• Achieves ≤ 5 seconds execution time per OCT cube on target hardware.
• Stable segmentation across scan variations and pathological cases.
• Clean, maintainable, and well‑documented C++ solution accepted at solution milestone.

• This role is responsible for the design, development, optimization, and validation of a non‑AI Retinal Nerve Fiber Layer (RNFL) and ILM segmentation algorithm for Swept‑Source OCT (SS‑OCT) data.
• The engineer will own the full technical lifecycle—from classical image‑processing algorithm design and prototyping to high‑performance C++ implementation using Intel IPP/MKL, ensuring accuracy, robustness, and runtime compliance on specified hardware.

 Key Responsibilities:

• Algorithm Design & Prototyping
• Design and implement classical (non‑AI) image processing algorithms to segment ILM and RNFL outer boundaries from SS‑OCT volumes.

Develop a robust segmentation pipeline using techniques such as:

• Noise reduction and filtering
• Edge detection and gradient analysis
• Thresholding and morphological operations
• Contour detection / boundary tracking
• Generate per‑A‑scan confidence scores (0–1) reflecting segmentation reliability.
• Ensure anatomical continuity and consistency of layer boundaries across neighbouring A‑scans.
• Prototype and validate algorithms using Python (NumPy/OpenCV) on diverse datasets, including healthy and glaucomatous cases.

 Optimization & Accuracy Tuning:

 Tune algorithm parameters to meet region‑specific accuracy requirements (ONH, fovea, peripheral, global).

• Validate segmentation accuracy against customer‑provided ground truth annotations.
• Handle variations in scan size (12×12 mm, 15×15 mm) and imaging conditions.
• C++ Solution Development & Performance Engineering
• Translate validated prototypes into production‑grade C++ code.

Optimize implementation using Intel IPP and MKL libraries for:

• SIMD/vectorization
• Multithreading
• Memory‑efficient processing of 3D OCT volumes
• Ensure execution time ≤ 5 seconds per OCT cube on Intel Core i5, 16 GB RAM reference hardware.

• Maintain functional parity between Python prototype and final C++ solution.

 Testing, Validation & Documentation:

• Support creation of algorithm‑level test cases and performance benchmarks.
• Assist in analyzing test results and resolving accuracy or runtime gaps.

Author or contribute to:

• Algorithm Design Documentation
• Developer/API documentation
• Build and configuration instructions