AI Engineer (Intern: Jan–May 2023 → Full-Time: Jan 2024–Present)
Virtusa Consulting Services Private Limited
– Completed GCP-Associate Cloud Engineer certification during internship; use Windsurf daily as a core part of engineering workflow for code generation, debugging, and code review – Designed and implemented multimodal RAG pipelines integrating OCR and Vision Models for document re-extraction; architected vector indexing and retrieval strategies that reduced manual validation effort by ∼70% – Built agentic AI workflows with tool calling, planning, and orchestration patterns to ingest Jira stories; implemented prompt orchestration and context assembly achieving ∼95% test case accuracy; reduced per-story test design time from 1–2 hours to under 5 minutes using AI-generated test cases – Designed multi-agent orchestration system with memory and planning patterns to monitor Jira story changes and automate re-triggering of documentation workflows and approval emails – Extracted requirements from PDFs and Figma using data ingestion and chunking strategies; modelled application screens as NetworkX Knowledge Graphs; implemented semantic search for test retrieval reducing manual test authoring by ∼65% per sprint – Built comprehensive RAG evaluation frameworks computing relevance, faithfulness, accuracy, and toxicity as KPIs; benchmarked RAG pipeline performance against industry alternatives using FastAPI/Django, and reported AI-assisted workflow ROI to stakeholders – Implemented hybrid search combining dense (FAISS) and semantic retrieval over indexed code repositories to automate API test script generation from .feature files; integrated LLM APIs (Claude, GPT-4, Gemini) into these services with attention to token limits, latency, and cost tradeoffs – Contributed to architecture decisions and technical design reviews for enterprise GenAI platform capabilities, including vector store infrastructure and retrieval optimization strategies; used Windsurf to build React and Node.js frontend interfaces for these tools, extending delivery across the full stack – Owned end-to-end delivery of AI-assisted features across design, build, and production support within Agile sprint cycles, collaborating directly with client engineering teams and communicating project updates to ensure stakeholder alignment – Contributed reusable RAG and agentic-orchestration components, prompt templates, and internal AI tooling standards to a shared knowledge base, accelerating onboarding and adoption of AI-assisted workflows across project teams
