Job Overview
We are looking for a senior C/C++ engineer to take an already-specified computer vision demo codebase and compile, integrate, and package it into a working macOS application (.app) for internal demo use.
This is not a research role and not a greenfield build.
The system architecture, kernel, CLI wrapper, and packaging blueprint are already defined. Your job is to:
• finalize build integration
• validate runtime behavior
• bundle dependencies
• produce a stable Apple Silicon .app demo package
The work is highly specified and should be completed quickly and precisely by the right engineer.
What You Will Be Working On:
You will be working on a deterministic computer vision demo system that:
• runs a fixed-point C kernel (no floating point math)
• processes raw video frames (.yuv)
• renders a 3-pane OpenCV visualizer
• streams live NDJSON telemetry to a terminal
• is packaged into a macOS .app for presentation to hardware partners
The core executable (cosimo_demo) already exists in specification form. Your task is to make it build, run, and package cleanly.
Scope of Work:
1. Build & Integration:
• Wire and validate CMake build:
• root CMakeLists.txt
• core/ (C deterministic kernel)
• cli/ (C++ wrapper)
• Ensure successful build on Apple Silicon (arm64)
2. Runtime Validation:
• Verify cosimo_demo runs correctly:
• frame ingestion from .yuv
• kernel execution (cosimo_compute_d0)
• 3-pane OpenCV visualizer
• quiver (vector) overlay rendering
• Validate CLI modes:
• standard run
• --headless
• telemetry rate control
3. Telemetry & I/O:
• Ensure strict telemetry contract:
• NDJSON output to stdout
• fflush(stdout) after each line
• no mixing of logs into stdout
• all diagnostics → stderr
• Validate latency measurement:
• must reflect kernel execution only
4. Fixture Validation:
• Confirm all demo fixtures:
• are 1920x1080
• luma8
• valid .yuv streams
• Ensure system fails cleanly on malformed input
5. macOS Packaging (.app):
• Execute packaging script:
• bundle executable into .app
• include fixtures
• bundle OpenCV dependencies (dylibbundler)
• Validate:
• Finder launch
• AppleScript launcher behavior
• fixture selection prompt
• terminal telemetry window
• visualizer window
6. Final Validation:
• Run full demo locally:
• verify all fixture modes
• confirm no crashes
• confirm telemetry + visuals behave correctly
• Deliver:
• working .app
• zipped distribution package
• brief build/run notes
What You Are NOT Doing:
• No ML model development
• No new algorithm design
• No architecture redesign
• No blockchain / backend work
• No UI/UX redesign
This is a precision execution and packaging task, not an exploratory project.
Required Skills:
• Strong C/C++ (5+ years)
• OpenCV (desktop apps)
• CMake build systems
• macOS development (especially Apple Silicon)
• Dynamic library handling (.dylib, install_name_tool, dylibbundler)
• CLI applications and debugging
• File I/O (binary / raw frame data)
Strongly Preferred:
• Experience packaging macOS .app bundles
• Experience with ARM64 environments
• Experience shipping internal tools or demos
• Experience debugging native runtime issues
Nice to Have:
• SIMD / low-level performance awareness
• Experience with video pipelines or frame processing
• Familiarity with deterministic or fixed-point systems
Project Timeline:
• Expected duration: 1 or 2 days, depending on how experienced and efficient you are.
• If you do a good job, we have more work for you. This is a discrete assignment to determine how efficient and effective you can be with instructions and objectives.
This assumes:
• codebase is already defined (it is)
• work is integration + packaging (not new development)
Deliverables:
• Fully working Apple Silicon .app
• Verified execution across all demo fixtures
• Clean telemetry output
• No runtime dependency issues
• Short documentation (build + run instructions)
How to Apply:
Please include:
1. Relevant examples of:
• C/C++ + OpenCV work
• macOS packaging or native apps
2. Confirmation you’ve worked with:
• CMake
• macOS .app packaging
3. Estimated time to complete (in hours)
Final Note
We are looking for someone who can:
take a well-defined system and execute it cleanly, quickly, and without introducing ambiguity
If you tend to over-engineer or redesign things unnecessarily, this is not the right fit.
If you are precise, fast, and experienced in native systems and packaging, this should be a straightforward and rewarding project.
ADDITIONAL CONFIDENTIAL SOFTWARE INSTRUCTIONS WILL BE GIVEN UPON HIRING.