VVR..

AI Engineer

Loading0%
BOOTING Open to AI Engineering roles

Vaibhav Vikas Ranjan Building intelligent systems for production

I’m an AI Engineer

Building agentic AI pipelines, LLM evaluation frameworks, and production ML tools that hold up under real-world messiness — not just clean benchmarks.

View My Work Resume
Vaibhav Vikas Ranjan
VVR
RAG Agentic AI AI Automation LLMs
Scroll
About

Turning research into
software that ships

Final-year Data Science & AI student at IIIT Naya Raipur, currently working as an AI Engineer at Accure Inc — moving between agentic reasoning research and the production systems that put it to work.

Agentic

Reasoning pipelines

Agentic AI workflows and prompt-engineering pipelines evaluated through a seven-dimension reliability framework.

Adversarial

Robustness testing

Stress-testing agentic systems against synthetic adversarial datasets to catch failure modes before production does.

Unsupervised

Anomaly detection

Detection systems trained on unlabeled data, built to catch defect types the training set never saw.

3

Roles held

4

Projects shipped

30+

Tools & frameworks

2026

Graduating

Toolkit

The stack behind
production AI

~25+ tools across the ML lifecycle — from first notebook to deployed pipeline.

Languages & Core

/ 05
Python SQL C++ JavaScript HTML/CSS

Generative AI & LLMs

/ 06
LangChain LangGraph MCP Chainlit Hugging Face Ollama

ML & Deep Learning

/ 06
PyTorch TensorFlow Scikit-learn OpenCV Pandas NumPy

Infra & Data

/ 08
Docker AWS FastAPI PostgreSQL ChromaDB Git Tableau Streamlit
// core_loop live

Daily orbit

tools_in_motion = 08 updated daily
Focus areas
Generative AI LLMs RAG NLP Computer Vision Anomaly Detection Data Engineering Prompt Engineering
Journey

From load-tested
systems to agentic AI

Three roles, each closer to the model layer than the last.

AI Engineer

Jan 2026 – Present

Accure Inc · Bangalore, India

  • Improved LLM reasoning reliability for AccureIQx by designing agentic AI workflows and prompt-engineering pipelines, evaluated through a seven-dimension framework.
  • Strengthened adversarial robustness of agentic reasoning systems, achieving high anomaly-detection accuracy across five stress-test scenarios using synthetic cybersecurity datasets.
  • Cut manual weekly reporting time by 35% by building automated Python analytics pipelines integrating multi-source EDA data via REST APIs.

Software Developer Intern

May 2025 – Jul 2025

Taskar — India’s 1st Healthcare Mall · Remote

  • Reduced manual reporting workflows by 40% via microservice-based RESTful API integrations for live healthcare and inventory data.
  • Improved analytics dashboard load times by 30% by optimizing UI rendering and async data fetching with React Hooks.
  • Increased user engagement 15% by redesigning dashboard layouts based on stakeholder feedback.

Software Engineering Intern

May 2024 – Jul 2024

Research Designs and Standards Organisation (RDSO) · Lucknow, India

  • Supported 500+ concurrent users at 99.9% uptime by engineering a responsive attendance tracking system (HTML, CSS, JS, PHP, MySQL).
  • Cut login times 60% for 1,000+ daily users via an optimized, AJAX-driven authentication workflow.
  • Validated production-grade performance under high concurrency using Apache JMeter load testing.
Selected Work

Research, shipped
as software

Agentic pipelines, vision systems, and NLP tools built end-to-end.

Agentic Contract Auditor 2026

CFR Compliance Checker

An agentic compliance analysis system that evaluates contract clauses against authoritative U.S. federal regulations using MCP-based retrieval, deterministic rules, LLM reasoning, evidence verification, durable memory, human review workflows, and production-ready persistence.

PythonFastMCP AgnoOllamaPydantic
status // operational View on GitHub
Multi-Agent Incident Response 2026

Distributed Agentic Incident & Security Response Engine

Kafka-based incident response combining a z-score detector with an LOF-based ML anomaly gate, a two-phase Investigator Agent over sandboxed MCP diagnostics, ChromaDB historical retrieval, and Slack/FastAPI human-approved remediation.

PythonAgno MCP / FastMCPPydantic v2 KafkaChromaDB FastAPIReact
status // operational View on GitHub
GenAI Career Copilot 2026

Smart ATS: Resume & JD Intelligence Platform

Explainable AI-powered ATS that uses LLM-based resume/JD intelligence, hybrid semantic + skill matching, RAG, LangGraph agents, and MCP tools for career insights..

PythonFastAPI LangGraphMCP PostgreSQLChromaDB KafkaStreamlit
status // operational View on GitHub
Churn Prediction & XAI 2026

Evidence-Grounded Telco Churn Prediction System

Evidence-grounded telco churn prediction using leakage-safe XGBoost, SHAP explanations, and a constrained LLM layer. Every LLM claim is deterministically verified against authoritative SHAP evidence (PASS/FLAG). The LLM is never the source of truth.

PythonXGBoost SHAPFastAPI OpenAIGradio MLflowDocker
status // operational View on GitHub
Issue → Verified PR Pipeline 2026

Multi-Agent Orchestration System

LangGraph pipeline (Researcher, Planner, Coder, TestWriter, Verifier, PRWriter) that turns a GitHub Issue into a verified Pull Request, with Docker-sandboxed verification and a mandatory human approval gate before any PR is opened.

PythonLangGraph DockerPydantic GitHub APISlack API
status // operational View on GitHub
Agentic Vision System 2026

Welding Anomaly Detection

Agentic industrial visual-inspection system built around sealed PatchCore anomaly detection. Uses deterministic anomaly routing, selective LLM reasoning, provenance-gated RAG over welding references, and MCP tools to produce grounded inspection reports with explainable heatmaps.

PyTorch PatchCore RAG MCP ChromaDB Docker
status // operational View on GitHub
Adaptive Security Evaluation 2026

AMDE — Adaptive MCP Defense Evaluation

Research framework for evaluating MCP/AI-agent security gateways, using bounded adaptive search to discover semantic bypasses and comparing them against static and budget-matched random baselines across real and replayed gateway targets.

Python MCP AI Security Docker Quality-Diversity Adversarial Testing
status // research complete View on GitHub
Contact

Let’s build something
reliable

Open to AI Engineering roles and interesting problems in agentic systems, RAG, or applied ML — reach out.

Start a conversation
Emailvaivikran@gmail.com LinkedInin/vaibhavvranjan GitHubvvXranjan