Job Description
Job Description:
We are seeking an experienced engineer who brings two distinct skill sets: AI/ML (Computer Vision) - owning the training, testing, and tuning of vision-based models for live camera-feed monitoring; and Native Android Development - developing and shipping a production Android app that talks to our cloud. You will partner with the deployment team to align models with their scaling/runtime constraints, work with the maintenance team to maintain and upgrade models.
Key Responsibilities
Design, train, evaluate, and tune computer vision models (detection, classification, segmentation, tracking) for live video and multi-camera use cases, including dataset curation, training/validation/testing pipelines, and rigorous benchmarking on accuracy, latency, and throughput.
Optimize models (architecture, quantization, pruning, distillation) to meet deployment and scaling constraints provided by the deployment team; re-tune or re-architect when those constraints change.
Partner with the operations team to maintain and upgrade models in production — triage regressions, refresh on new data, address drift, and ship improved versions.
Design, build, and ship a native Android application (Kotlin, Jetpack Compose, MVVM/Clean Architecture, Hilt, Coroutines/Flow, Room, WorkManager) that interacts with our cloud backend.
Build secure cloud integration: REST/gRPC, OAuth2/JWT, TLS, FCM push, and offline-first sync; handle Android runtime permissions and background execution correctly.
Set up Android CI, testing (unit + instrumentation), crash reporting, and Play Store release pipelines.
Required Qualifications
5+ years of total professional software engineering experience.
Proven experience in vision-based model training and testing, with models shipped to production.
Hands-on experience with cloud-based AI training/experimentation infrastructure (AWS SageMaker, GCP Vertex AI, Azure ML, or equivalent).
Experience building models for continuous monitoring via live video input from cameras and tuning them for efficient inference across multiple concurrent camera streams.
Track record of collaborating with deployment/MLOps/operations teams - translating runtime constraints into model decisions and supporting models post-launch.
Proficiency in Python and ML frameworks (PyTorch, TensorFlow, TensorFlow Lite, ONNX); strong grasp of CNNs and modern detection/tracking architectures (YOLO, DETR, ByteTrack, etc.).
Strong proficiency in Kotlin and the modern Android stack (Jetpack, Compose, Coroutines/Flow, Hilt, Room, WorkManager); demonstrated experience shipping production Android apps to the Play Store.
Deep working knowledge of Android Wi-Fi and BLE APIs.
Experience with cloud connectivity on Android (REST, gRPC, WebSockets, or MQTT; OAuth2/JWT; TLS; FCM) and Android’s security/permissions model (runtime permissions, foreground services, background execution limits).
Comfort interfacing with embedded/IoT hardware over BLE and WiFi under real-world conditions (intermittent connectivity, retries, power constraints).
Preferred Qualifications
On-device inference acceleration (Android NNAPI, Qualcomm SNPE, MediaPipe, GPU delegates).
Streaming protocols (RTSP, WebRTC, HLS) and video codecs; edge / IoT camera deployments.
Model monitoring, drift detection, and active learning loops.
Contributions to open-source ML or Android projects.
Education
Bachelor’s or Master’s degree in Computer Science, AI/ML, or related field — or equivalent practical experience.
measurement bias
Strong product mindset and bias for measurement - you instrument, benchmark, and optimize.
Excellent collaboration skills with deployment and operations teams, and a maintenance mindset toward models in production.
Ability to switch effectively between two distinct domains (ML model development and Android app development) and deliver in both.