Backend & Full-Stack Engineer - GCP & AI

SAMRIDHI THAKUR

Software Engineer Backend · Full-Stack · Applied AI · Cloud (GCP)

Immediate Joiner Open to SDE 2, Backend, Full-Stack, AI/ML and Cloud roles Gurugram / Delhi NCR

+91-935******* *****************@*****.*** linkedin.com/in/samridhi-thakur-764aa8203 PROFESSIONAL SUMMARY

Software engineer with 3+ years of experience building and supporting production healthcare systems on GCP: FastAPI/Flask services, REST APIs, WebSocket real-time workflows, React and Salesforce LWC front ends, and LLM-powered question answering (RAG at the API and prompt layer). Contributed to Advocate Assist (400K–500K advocates onboarded) and a Member Summary workflow used by ~15,000 CRM users. Strong in production root-cause analysis across application, authentication and cloud boundaries; maintained 15+ Terraform modules; co-led a team of 10+.

TECHNICAL SKILLS

Languages: Python, JavaScript, SQL/SOQL, Apex, Java, C++ Backend & APIs: FastAPI, Flask, Node.js, REST API Design & Versioning, WebSockets, Microservices, Event-Driven Systems, Redis, JWT Authentication

Frontend: React, Salesforce Lightning Web Components (LWC), Angular, HTML, CSS AI/ML & LLM: LLM-Integrated Applications, RAG (API/prompt layer), LLM Answer Evaluation, Embeddings, Vertex AI, Semantic Similarity

(Cosine, Jaccard), Entity Resolution, Transfer Learning, Knowledge Distillation Cloud & DevOps: GCP (Cloud Run, Pub/Sub, BigQuery, Cloud Logging, Cloud Monitoring), Terraform (IaC), CI/CD, Nginx, Git Engineering: System Design, Distributed Systems, Production Debugging, Root-Cause Analysis, Code Reviews, Technical Planning, Mentorship PROFESSIONAL EXPERIENCE

Associate Software Engineer II — Optum (Healthcare Technology), Gurugram, India Sep 2023 – Oct 2026

• Architected a semantic matching and entity-resolution pipeline (cosine and Jaccard similarity, substring signals, custom tie-breakers) to reconcile inconsistent upstream benefit metadata, unblocking a release-critical feature for ~15,000 CRM users without waiting on upstream API fixes.

• Integrated multiple Member Summary APIs with correlation logic for inconsistent upstream responses, automating summary generation that was previously written and sent manually, in a workflow used across ~15,000 CRM users.

• Built a question-blocking feature for an LLM-powered Q&A system (block by user, app or source document) with an audit schema capturing actor, action and change history on every CRUD operation, eliminating known inaccurate answers and adding traceability.

• Designed cross-session Q&A persistence across React and Salesforce LWC widgets: history and feedback retrieval APIs, Redis/BigQuery feedback path and CRM-specific session storage, letting users restore conversations after reloads; zero reported defects after release.

• Led the migration of proactive-question workflows from Salesforce Platform Events to WebSockets after daily usage exceeded its allocation, eliminating Platform Event costs and improving response time by ~0.5–1 s; earlier cut subscription load by mapping users across multiple event channels with a weekly scheduled sync job.

• Diagnosed recurring WebSocket disconnects (close codes 1001/1005) with transport-level logging, tracing them to duplicate heartbeat handling after a server migration; re-engineered connection lifecycle (module-level connection, grace-period timeout, call-ID mapping) to stop reconnects on UI reloads across 3 CRM integrations.

• Root-caused production issues across service boundaries: a CRM connection failure (JWT secret and signature mismatch after a FastAPI migration, solved without direct CRM access) and Cloud Run unhealthy instances with high latency (Vertex AI embedding call returning HTTP 429 at startup).

• Maintained and modified 15+ Terraform modules across 4 project/environment configurations on GCP; reviewed infrastructure changes and deployed through CI/CD (Cloud Run, Pub/Sub, Cloud Logging, Cloud Monitoring).

• Built a Node.js WebSocket test client that cut connection-test setup from ~10 minutes to ~1 minute (~90% faster) and a React-to-Nginx- to-Cloud Logging path for frontend logs, with no new service or Terraform module.

• Led technical planning and component architecture for a shared React widget intended for 3 CRMs (estimates, grooming, QA coordination); wrote workflow documentation that eased tool adoption and new-developer onboarding.

• Co-led a team of 10+ engineers through solutioning and PR reviews; mentored 2 engineers and unblocked teammates on features and production issues.

SDE Intern — Optum Jun 2022 – Aug 2022

• Built Angular front-end components for internal healthcare workflows; designed wireframes and UX flows in Figma. Research Intern — National Institute of Technology Delhi Jun 2021 – Jul 2021

• Built an IoT plant-disease detection prototype with sensor-driven cloud data transfer, real-time monitoring and Django front-end components.

PROJECTS

• Histopathological Image Classification: Built a breast-cancer image-classification model using transfer learning and knowledge distillation, improving performance through feature-extraction and model-optimization experiments.

• Trouvaille, Travel Service Platform: Developed backend workflows and MySQL database integrations for a full-stack travel platform in a 3-member team.

EDUCATION & CERTIFICATIONS

B.Tech, Computer Science — National Institute of Technology Delhi 2019 – 2023 Certification: Salesforce Platform Developer I

Contact this candidate