Data Scientist - Computer Vision & Machine Learning

Job Summary

The Automatic Number Plate Recognition (ANPR) team is building a next-generation Edge-to-Cloud machine learning pipeline to power Multi-Lane Free Flow (MLFF) tolling across Indian highways. As a Data Scientist, you will play a critical role in scaling our architecture to handle high-speed traffic, extreme environmental edge cases (mud, glare, tampering), and highly localized typography. You will take ownership of model training, failure mode diagnostics, and the active learning ecosystem that bridges our lightweight Edge models with massive Cloud Vision-Language Models (VLMs).

Job Type

Full-Time

Responsibilities

• Model Development & Fine-Tuning: Train, optimize, and evaluate state-of-the-art object detection, Pose Estimation (for License Plate Normalization), and Optical Character Recognition models for high-speed, noisy highway environments.
• Failure Diagnostics & Data Curation: Analyze large-scale, unstructured video data from tolling gantries. Utilize dimensionality reduction and advanced clustering techniques on high-dimensional vehicle embeddings to mathematically isolate and diagnose visual failure modes.
• Active Learning & AutoGT: Build and maintain programmatic feedback loops using Cloud VLMs to automate ground truth generation (AutoGT) and dynamically retrain edge models based on real-world anomalies.
• Tracking & Heuristics: Implement and refine temporal tracking logic and spatial prior mathematics to stabilize OCR predictions and handle object occlusion.
• Multimodal Fusion Readiness: Collaborate on architecting feature embeddings that fuse 2D visual data with physical sensor telemetry (LiDAR profiling, Radar speed).
• Edge Optimization: Collaborate with deployment engineers to ensure models adhere to strict latency and compute constraints for real-time edge hardware execution.

Disclaimer: This job posting has been aggregated from external source. Role details, content, and availability are subject to change. Applicants are advised to confirm the latest information directly on the company website before applying.