Ajinkya Joshi

Machine Learning Engineer (Computer Vision) with 2+ years of hands-on experience designing and deploying end-to-end AI solutions for traffic analytics, surveillance, ANPR systems, and geospatial intelligence applications. Skilled in developing scalable object detection, OCR, and video analytics pipelines using YOLOv8, OpenCV, and deep learning frameworks. Experienced in model training, optimization, inference acceleration, and production deployment for government-scale traffic enforcement systems.

AI/ML Engineer (Computer Vision Intelligent Systems) at INAI IIIT Hyderabad (2025-01 – Present)

• Trained and fine-tuned custom YOLOv8 detection models on DDPAI X2S Pro dashcam datasets tailored for Indian road, traffic, and surveillance scenarios.
• Engineered intelligent vehicle movement analysis pipelines for wrong-side driving detection using centroid trajectory estimation and automated tracking mechanisms.
• Built a real-time dashcam-driven traffic monitoring solution in collaboration with Hyderabad Traffic Police for identifying violations from moving patrol vehicles.
• Implemented automated number plate recognition workflows using YOLOv8, PaddleOCR, and regex-based verification for accurate vehicle identification and digital evidence creation.
• Improved license plate extraction reliability under motion blur, nighttime footage, and challenging environmental conditions for enhanced OCR performance.
• Developed automated violation evidence generation modules contributing to challans worth Rs. 3 Lakhs+ within 10 days of deployment in the non CCTV servilance zone in Old Hyderabad area.
• Accelerated large-scale inference workloads through multi-GPU parallel processing, reducing execution time from 3X to nearly 1.25X for near real-time analytics.
• Designed and deployed a Flask-based enforcement platform integrated with Telangana Government APIs for seamless transmission of detected violations to the Telangana Police database for challan processing.
• Created operational dashboards for live validation workflows, rejection analysis, enforcement monitoring, and challan revenue tracking with historical reporting capabilities.
• Successfully deployed pilot traffic enforcement solution with Hyderabad Traffic Police, generating challans worth Rs. 20 Lakhs during a 20-day operational deployment.

Machine Learning Intern at INAI IIIT Hyderabad (2024-01 – 2025-01)

• Developed multilingual AI-driven safety video translation pipeline supporting 11 Indian languages using Google Speech-to-Text, Bhashini API, gTTS, and FFmpeg.
• Built aerial imagery analysis workflows for geospatial change detection and land-use classification using drone surveillance data.
• Performed accident blackspot identification, segmented area analysis, and ArcGIS visualization for NHAI road safety initiatives.
• Integrated spatial intelligence and analytics modules for infrastructure monitoring and geospatial decision-support applications.

Bachelor of Technology (B.Tech) in Information Technology – St. Vincent Pallotti College of Engineering and Technology, Nagpur