Spatiotemporal Analytics Engineer (Edge) Junior

Role Overview: As a member of the team based in Hyderabad, India, your role will involve developing real-time detectors for diversion, contraband, and bullying by utilizing a blend of statistics, geometry, and lightweight machine learning on an NVIDIA-accelerated edge box. Your primary responsibility will be to transform raw tracks and poses into understandable pattern scores while meeting stringent latency constraints. Key Responsibilities: - Design baseline zones and lead-lag signals (such as TLCC, CUSUM/BOCPD, LombScargle, DTW) to identify coordinated decoys. - Utilize a combination of geometry and pose cues to encode encirclement, follow, and throw actions, and produce JSON outputs rich in evidence. - Develop vectorized, well-tested Python code (leveraging NumPy/PyTorch) capable of executing in less than 50 ms per window on the Box. - Collaborate with the DeepStream/TensorRT pipeline to handle atomic events and generate pattern alerts. Qualifications Required: - Demonstrated proficiency in algorithms, probability, time series analysis, and change detection. - Strong command over Python 3.11, NumPy, Pandas, basic PyTorch, and experience in performing unit tests and benchmarks. - Ability to analyze latency, precision/recall metrics, and make informed decisions regarding tradeoffs between false positives and negatives. If this opportunity interests you and you are excited about contributing to detectors that directly impact individual safety, please forward your cover letter and resume to Ashwin at ashwin.jaiswal@safespaceglobal.ai.,