AI Engineer with 6+ years in Computer Vision, OCR & Generative AI
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
Machine Learning Engineer with 5+ years of production experience in Computer Vision, OCR systems, and Generative AI. Shipped deep learning pipelines achieving up to 95% extraction accuracy and eliminating 70% of manual data-entry workflows. Architected a production grade agentic AI refund system using LangGraph stateful orchestration, FastAPI async endpoints, tool calling agent loops with fraud detection, and prompt injection defense deployed end-to-end with Docker. Experienced in LLM powered RAG applications using LangChain, FAISS, and HuggingFace, with GPU optimised inference pipelines reducing latency by 35%.
Bharathidasan University
Master of Computer Applications (MCA)
August 1, 2017 – June 30, 2019
Shrimati Indira Gandhi College
Bachelor of Computer Applications (BCA)
August 1, 2014 – June 30, 2017
DSM Soft Pvt Ltd
Machine Learning Engineer
December 1, 2019 – Present
India
AI Refund Agent - End-to-End Agentic AI Refund Automation Platform
January 1, 2026 – June 1, 2026
• Architected end-to-end agentic AI refund system using LangGraph stateful graph (StateGraph + RefundState) with 6 nodes (extract, lookup, policy, decide, auto process, escalate) replacing a raw function calling loop with production-grade stateful orchestration. • Implemented fraud detection flagging customers with 3+ refund requests within 90 days, and built prompt injection defence blocking 100% of policy override attempts in the test suite. • Built streaming FastAPI backend with Server Sent Events delivering real time agent reasoning steps and tool call traces to Streamlit frontend. • Deployed full system with Docker Compose provider agnostic LLM integration supporting Groq, OpenAI, and Anthropic APIs without code changes. • Designed decision engine enforcing 5 refund scenarios — approved, denied (final sale / digital / outside 30-day window), and escalated (>$500) — across 20 edge case test orders. • Production upgrades in progress: Redis + Celery async agent pipeline, JWT auth + Redis rate limiting, LangSmith step-level observability, 20-case LLM-as-judge evaluation suite targeting 94%+ accuracy, Kubernetes autoscaling for Celery workers.
View ProjectRAG Chatbot - PDF Question Answering System
January 1, 2026 – June 1, 2026
• Engineered end-to-end RAG system using LangChain, FAISS vector search, and HuggingFace sentence transformer embeddings for semantic querying over large PDF corpora. • Improved answer relevance over naive LLM prompting using recursive character chunking, retrieval-grounded context injection, and hallucination-prevention prompting. • Deployed FastAPI-backed chatbot with Streamlit frontend supporting real-time PDF ingestion, chunking, and semantic search querying.
Cultural Fit Analysis
The candidate's project portfolio demonstrates a strong alignment with an AI Engineer role, particularly in agentic AI, generative AI, and computer vision. The diversity of projects, from OCR pipelines to RAG chatbots and agentic systems, shows a broad interest and capability in different AI domains. The independent nature of the AI Refund Agent and RAG Chatbot projects suggests initiative and self-direction, which are valuable traits for cultural fit in innovative teams.
Soft Skills & Operational Fit
The candidate's resume highlights a strong focus on problem-solving and delivering measurable business impact (e.g., 'eliminating 70% of manual data entry', 'reducing manual review time by 30%'). The detailed project descriptions suggest an ability to work independently and drive projects from architecture to deployment. The mention of 'production upgrades in progress' for the AI Refund Agent project indicates a proactive and continuous improvement mindset.