Computer Vision Engineer

Hackajob
Hackajob

Software Engineering

United States

Posted on Aug 5, 2026
hackajob is collaborating with Leo Technologies to connect them with exceptional professionals for this role.

  • Build and fine-tune custom object detection, segmentation, and tracking models and related architectures for client- or product-specific use cases.
  • Design data pipelines covering collection, annotation (Roboflow, CVAT, Label Studio), augmentation, and curation to produce training-ready datasets.
  • Train and evaluate models using PyTorch, optimizing for mAP, precision/recall, and latency targets specific to each deployment.
  • Optimize models for inference using TensorRT, ONNX, and quantization (FP16/INT8) to hit real-time performance on target hardware.
  • Integrate vision models into broader systems through REST/gRPC APIs, video stream processors (RTSP, GStreamer), or edge runtimes.
  • Monitor production models for drift, edge cases, and failure modes; build retraining loops to keep performance high over time.
  • Collaborate with product, hardware, and software teams to scope problems, set realistic accuracy/latency targets, and ship.

Requirements

What You Bring To The Table

  • 3+ years of hands-on computer vision experience with at least one production deployment of a custom-trained detection or segmentation model.
  • Strong Python skills and fluency with PyTorch; comfortable reading and modifying model code, not just calling high-level APIs.
  • Working knowledge of the NVIDIA stack: CUDA fundamentals, TensorRT, and at least one of DeepStream, Triton, or Jetson deployment.
  • Experience with the full data lifecycle: sourcing imagery/video, designing annotation schemas, and handling class imbalance and edge cases.
  • Solid grasp of evaluation metrics (mAP, IoU, confusion matrices) and what they actually mean for a given application.

Nice to Have

  • Experience with multi-object tracking (ByteTrack, BoT-SORT, DeepSORT).
  • Background in pose estimation, OCR, or 3D vision.
  • Familiarity with synthetic data generation (Omniverse Replicator, Unity Perception).
  • MLOps experience: experiment tracking (W&B, MLflow), CI/CD for models, containerization (Docker).
  • Domain experience in security / law enforcement / military / intelligence
  • Contributions to open-source CV projects.