Computer Vision + RAG + AI Systems
AdityaRajesh
Building production-grade AI systems across Computer Vision and Retrieval-Augmented Generation, with focus on reliability, evaluation, and real-world deployment.
Client Validation+ Credentials
Documented results, client validation, and third-party feedback from real deployments.
FPT Software
Mid-Journey Stories: FPT Global Interns Driving Growth & Innovation
"The program strengthened my Computer Vision foundations and practical decision‑making across production ML workflows."
Contributed to an AI-powered inspection system for automotive manufacturing, focusing on model evaluation, data quality, and deployment readiness.
Client Acceptance & Validation
Official project completion verification
Client Acceptance Letter
Documented completion of engineering deliverables
Client
Hyundai Kefico · Hanoi Branch
Date
September 2025
Scope
Official client acceptance for an AI-powered defect identification system supporting detection and segmentation in automotive manufacturing.
Professional Certifications
Industry-recognized credentials and achievements
Certificate of Completion — Global Internship Program (2025)
FPT Software
Completed a full-time internship focused on Computer Vision and Retrieval‑Augmented Generation (RAG) systems for production-grade inspection workflows.
Endorsed by Global Collaborators
Recommendations from mentors and team members across time zones
Quynh Anh Nguyen
MA Strategic Marketing | Marketing Officer
FPT Software
"I had the pleasure of observing Aditya Rajesh during the TalentSphere – FPT Global Internship 2025. As a team leader, he showed great adaptability and collaboration, uniting teammates across time zones and inspiring them to deliver impactful results. A true future global leader!"
View on LinkedInAll recommendations verified through LinkedIn connections
Selected Projects
A focused set of engineering projects with measurable outcomes and production-grade delivery.
Deep Learning-Based Visual Inspection System
Industry-grade AI inspection system for Hyundai using YOLOv8, OpenCV, and Python, supporting detection and segmentation workflows.
- +Aligned model performance with industrial requirements through iterative evaluation
- +Coordinated technical sync-ups and sprint reviews with stakeholders
- +Communicated experiment results, dataset issues, and model limitations
- +Delivered a production-ready inspection system with client acceptance
AI-Powered Footfall Counter with Zone-Based Tracking
Production-grade video analytics system using YOLOv8 for real-time person detection with polygon-based zone tracking across multiple environments. Implements FastAPI backend with custom inference pipelines optimized for accuracy and throughput.
- +Multi-zone polygon tracking with 7 distinct zones for complex crowd analysis
- +Frame-by-frame inference at 1280px resolution with confidence thresholding (>0.5)
- +Supervision library integration for zone triggers and annotated video generation
- +RESTful API with multipart upload, progress tracking, and automatic file cleanup
- +Handles Market Square (multi-zone), Grocery Store (single ROI), and Subway (corridor) scenarios
TMFeye AI — Continuous Trial Master File Intelligence Platform [Upcoming]
Enterprise AI platform for pharmaceutical compliance leveraging production-grade RAG architectures. Delivers continuous Trial Master File oversight with contextual document analysis, narrative risk assessment, and real-time health scoring for regulatory inspection readiness.
- +Production system under active development for global pharmaceutical sponsors
- +Advanced retrieval-augmented generation pipeline for regulatory document analysis
- +Contextual understanding engine with root cause identification and impact assessment
- +Real-time TMF health scoring with proactive gap detection and mitigation workflows
- +Enterprise deployment architecture targeting 80% reduction in filing time and 99.5% document accuracy
Let's build reliable AI systems.
Open to roles and collaborations in Computer Vision, RAG, and applied machine learning.