Senior/VP CV/AI/ML Engineer for video analysis

📡 Senior Computer Vision Engineer - Scalable Video Intelligence Systems

Location: Hybrid (Israel preferred)

Type: Full-time

Seniority: Senior / Principal / Founding Engineer

The Mission

We’re building AI systems that watch the real world at scale - intelligently, securely, and fast. This is not just another CV startup. We're working with real-time video at the edge, deploying state-of-the-art vision models across distributed environments, and pushing the boundaries of what LLMs + VLMs can do when they meet massive, high-frequency sensor data.

This role is NOT for someone who follows the frontier of AI – bur rather lead and create it.

What You’ll Do

• Design and train VLMs (CLIP, BLIP, LLaVA-style) for use cases like zero-shot detection, re-identification, tracking, open-vocabulary search, and scene understanding.

• Implement high-throughput, low-latency inference pipelines using tools like TensorRT, DeepStream, or custom CUDA kernels.

• Build scalable tooling to extract structure, features, metadata. and meaning from high-frequency visual data from texts/prompts .

• Build efficient reasoning mechanisms for video understanding that runs in low compute environments 

• Integrate distributed and redundant CV systems in multi-tenant, secure environments - where failover, logging, and observability matter.

• Architect and deploy real-time video analytics systems across edge/on-prem environments - including camera ingestion, inference pipelines, and smart alerting.

• Develop internal tools for video search and semantic indexing, and real-time event tagging over massive, continuous video streams.

• Collaborate with infrastructure engineers, security specialists, and applied AI researchers to ship scalable, production-grade systems.

The Perfect Candidate

• Lives and breathes the most cutting-edge research and tools in AI - including the latest in vision-language models, detection/tracking systems, and multi-modal workflows.

• Has designed and architected surveillance-grade CV systems capable of managing thousands of concurrent video streams across on-prem and edge nodes.

• Has trained models for similarity search and scene classification, or multi-modal alignment.

• Built systems combining different VLMs and LLMs for real-time reasoning, scene understanding, or explainable AI in vision-based workflows.

• Stays on top of arXiv drops but is a ruthless implementor - focused on real-world performance, not academic vanity. No loose screws. Every system runs clean.

• Thinks like a systems engineer, executes like a hacker, and debugs like a surgeon.

• Thrives in ambiguity and is obsessed with stability, latency, and observability at scale.

Tech Stack & Environment

- Languages: Python, CUDA, C++, Go (bonus)

- Frameworks: PyTorch, TensorFlow, OpenCV, MMDetection, Detectron2

- AI/ML: CLIP, LLaVA, BLIP, YOLOv8, SAM, Segment Anything, DeepSort, Transformer-based tracking/detection models

- Tooling: TensorRT, DeepStream, ONNX, Triton Inference Server, Weights & Biases

- Infra: Kubernetes, Docker, gRPC, Kafka, Redis, NVIDIA Jetson/Orin, edge compute

- Storage & Indexing: Chroma, FAISS, Milvus

You Might Be a Fit If You Have:

- 5+ years in AI/ML or CV-heavy production environments

- Proven ability to ship real-time CV pipelines (on-prem, edge, or low-latency cloud)

- Deep familiarity with embedding spaces, contrastive learning, detection/tracking models

- Experience deploying systems where accuracy, latency, and uptime are equally critical

- A hacker/researcher mentality: you can implement papers quickly, then optimize the hell out of them

Bonus Points For:

- Experience in multi-camera synchronization, thermal/IR models, or video anomaly detection

- Building CV/ML systems for regulated industries (e.g. defense, smart cities, industrial monitoring)

- Contributions to open-source AI projects (or publications in CVPR, ECCV, NeurIPS)

- Understanding of security/zero-trust principles in sensitive deployments

Why Join Us

- You'll be one of the first vision engineers on a team building a global intelligence layer on top of the physical world.

- You'll have direct impact on millions of lives, and on some of the most well-known and high-profile entities in the world

- Work on the frontier - LLMs, multi-modal reasoning, distributed systems, and real-time vision all come together here.

- No hype - just execution - We care less about the buzzwords and more about what works in the field, at scale.

- Mission-driven - Our work improves safety, awareness, and decision-making in the environments that need it most.