Rajdeep Thaosen
Backend engineer · Java · Spring Boot · GCP
Backend Engineer with a Master's in Computer Science from IIT Kharagpur. Production experience in Java, Spring Boot, and Google Cloud Platform building regulated eKYC APIs at a fintech registrar. Led a 3-engineer AI team that shipped a computer vision system to the Indian Ministry of Defence.
Experience
Jul 2025 — Present
CAMS, Chennai
Platform Modernisation Program (CAMSKRA)
Software Development Engineer, Backend
KYC Hub — investor and client-facing eKYC platform
eKYC APIs. Build and maintain Spring Boot REST APIs in Java for investor-facing eKYC flows and client-facing Push APIs consumed by Asset Management Companies (AMCs); contracts documented via OpenAPI / Swagger; shipped in Scrum on GCP with AlloyDB managed PostgreSQL.
Partner onboarding. Onboarded PhonePe and Doctor Finance end-to-end: Apigee apps with tenant credentials, Apigee-issued JWT auth, and RBAC across 10+ roles for the client / partner / agent hierarchy behind In-Person Verification (IPV).
Aadhaar eSign. Integrated HyperVerge APIs into the KYC journey, handling partner auth, retries and regulatory compliance for a flow consumed across multiple AMC integrations.
Deployment and integration owner — three internal applications
Enterprise Delivery Hub — company-wide initiatives dashboard
Team Odin — internal RBAC service
SyncPulse — daily standup app with log capture, a log dashboard and history
GKE delivery. Integrated and deployed all three onto Google Kubernetes Engine via GitLab CI/CD pipelines, Helm charts and ArgoCD GitOps; debugged inter-service communication across cluster DNS, Apigee proxies and Cloud Logging; serving ~200 stakeholders across 12 roles.
Enterprise SSO. Implemented Microsoft Entra ID (Azure AD) SSO end-to-end — app registration through token validation and RBAC; coordinated delivery with Infra, DevOps and InfoSec.
Aug 2024 — May 2025
Salphan Energy Pvt. Ltd., Odisha
Industry-sponsored project for Ordnance Factory Badmal, Ministry of Defence
AI Team Lead, Computer Vision
Team lead. Led a 3-engineer AI team replacing the factory's manual inspection of critical manufacturing equipment for cracks, piping and cavities with an end-to-end computer vision system.
DICOM pipeline. Built the pre-processing pipeline (PaddleOCR for serial / batch / orientation metadata) plus upstream classification models for outlier filtering and orientation correction before the anomaly detector.
Data. Owned annotation and validation of 25,000+ DICOM images across multiple anomaly classes; set the labelling protocol used by the team.
Models. Trained and benchmarked YOLOv8, YOLOv11 and RT-DETR for multi-class anomaly detection on industrial X-ray imagery; YOLO variants outperformed RT-DETR and were chosen for deployment.
Outcome. ~85% precision on on-site validation by end of tenure; client adoption led to a follow-on project for the company.
Projects
Personal project · Apr — Jun 2026
github.com/rajdeepthaosen7/veriKYCVeriKYC
CV-powered KYC verification platform
Stack. Java Spring Boot orchestration, FastAPI (Python) ML inference tier, React frontend — a production-shaped eKYC stack across PAN, Aadhaar, Passport, Driving License and Cheque.
Inference. EfficientNet-B0 (ONNX) document classification in the synchronous Spring Boot request path via WebClient; PaddleOCR field extraction with per-document validation (PAN regex, Verhoeff checksum, IFSC, MRZ parsing).
Deploy. GCP Cloud Run with Cloud SQL (PostgreSQL); independently deployable microservices with JWT auth, RBAC and paginated document management APIs.
M.Tech thesis · IIT Kharagpur
CV for industrial X-ray anomaly detection
with Ordnance Factory Badmal
Multi-class anomaly detection on industrial X-ray imagery — the research basis for the Salphan deployment above.
Skills
Education
M.Tech, Computer Science & Data Processing
Indian Institute of Technology, Kharagpur · July 2023 – June 2025 · CGPA 8.01 / 10
Thesis: CV for Industrial X-ray Anomaly Detection — with Ordnance Factory Badmal