Hi, I'm Pranshu Ghori
AI Engineer · LLM Agents
I build AI systems that ship — agentic RAG pipelines on Azure, LLM-powered workflows, and the production data infrastructure on AWS that feeds them.
About Me
AI engineer shipping production-grade agents, RAG systems, and the data pipelines behind them.
I'm pursuing concurrent degrees in B.S. Artificial Intelligence (STEM) and B.S. Business Data Analytics (STEM) at Arizona State University, maintaining a 4.0 GPA. My focus sits at the intersection of language models, autonomous agents, and real-world data systems.
Currently, I'm a Data Science Co-Op at Command Credit Corp, building production data pipelines on AWS serverless infrastructure (Lambda, SST) and shipping statistical risk models over a 72M-record dataset — including a distribution-driven classification scheme now used for risk segmentation by the credit underwriting team.
I specialize in LLM engineering, agentic orchestration with LangGraph, and RAG pipeline design. From building multimodal compliance auditing systems on Azure — using Azure Video Indexer, Azure AI Search, and GPT-4o — to production RAG pipelines over FAISS and Pinecone, I engineer AI systems that reason, retrieve, and act with full observability via LangSmith and Azure Application Insights.
I'm particularly drawn to autonomous AI agents, domain-specific tooling, and applied generative AI — systems where language models move beyond chat and become reliable, observable components in production workflows. My data science and ML engineering background grounds the AI work in statistical rigor and solid engineering fundamentals.
LLM Engineering
Prompt engineering, API integration, and chaining LLMs into reliable, production-grade pipelines.
Agentic Systems
Multi-agent orchestration with LangGraph — planning, tool use, reflection, and conditional routing.
RAG & Retrieval
Vector search, embeddings, and knowledge retrieval using FAISS, Pinecone, and Azure AI Search.
Skills & Technologies
The stack I use to design, build, and deploy intelligent AI systems.
Agentic AI & LLM Engineering
Machine Learning & Deep Learning
Software Engineering
Data & Visualization
Experience
Production data infrastructure, ML engineering, and applied AI — systems that ship.
Data Science Co-Op
- Build and deploy production data pipelines on AWS serverless infrastructure (Lambda, SST) for CommandInsight — an internal risk analytics platform powering the credit underwriting team.
- Own distributional analysis of a financial risk index across a 72M-record production dataset — identified a right-skewed, gamma-distributed structure and shipped a three-tier classification scheme now used for risk segmentation.
- Run fill-rate analysis across 72M records of production data feeds to quantify field-level completeness and surface data quality issues before they reach downstream models.
Data Analytics Assistant
- Analyzed network hardware inventory data in Python, Pandas, and SQL across 1,000+ assets — identifying discrepancies, null fields, and duplicate records to improve data quality for audit cycles.
- Built automated ETL pipelines to clean, reconcile, and visualize inventory datasets using Python and Power BI — eliminating manual reconciliation steps and delivering recurring dashboards to infrastructure leadership.
Web Assistant
- Maintained 20+ university web pages using HTML and CSS; performed pre-publication QA validation catching layout breaks, dead links, and responsive failures before production.
ML Engineer Intern
- Assisted the ML engineering team building end-to-end production ML pipelines — contributing to data preprocessing, feature engineering, and model training workflows using Python and Scikit-learn.
- Built and tested classification and regression models using Scikit-learn, supporting model selection, hyperparameter tuning, and performance benchmarking across multiple production use cases.
- Supported MLOps workflows by contributing to model deployment and monitoring tasks — gaining hands-on experience with production-grade ML lifecycle management and pipeline integration.
Featured Projects
AI agents, RAG systems, and data pipelines — built end-to-end.
Video Compliance QA Pipeline
- Built a production-grade video compliance auditing system orchestrated by LangGraph — ingesting multimodal content via Azure Video Indexer (transcripts + OCR) and detecting regulatory violations using RAG powered by Azure AI Search and Azure OpenAI embeddings.
- Engineered the core reasoning engine using GPT-4o to deterministically synthesize compliance rules against extracted video content, generating structured JSON reports.
- Integrated LangSmith for LLM tracing and Azure Application Insights for production-grade telemetry and full-stack observability.
- Designed end-to-end modular architecture with clean separation across ingestion, retrieval, reasoning, and reporting stages — deterministic outputs with deep observability at every layer.
Corporate Brochure Generator
- End-to-end pipeline that scrapes any corporate website and generates a polished company brochure
- Fetches all links from a homepage, then uses Grok to filter only relevant pages (About, Products, Careers, etc.)
- Scrapes each selected page and passes consolidated content to Grok for brochure generation
- Outputs formatted Markdown ready for publishing or export
- Modular structure: scraper.py handles all web utilities, brochure.ipynb orchestrates the full pipeline
DocumentLoader — Production RAG Pipeline
- Built a production-ready RAG pipeline supporting natural language Q&A over custom documents with streaming responses, source citations, and multi-turn memory — reducing retrieval latency via FAISS nearest-neighbor search and dynamic document ingestion at runtime.
- Orchestrated the full retrieval-to-reasoning loop using LangChain and LangGraph — context retrieval, prompt construction, LLM chaining, and fallback handling.
- Architected for full provider portability (FAISS → Pinecone, OpenAI → Anthropic) without re-architecting the pipeline.
California Housing Price Prediction
- Built an end-to-end regression ML pipeline
- Used stratified train/test split, preprocessing with ColumnTransformer, and unified Pipeline
- Engineered predictive features like log transforms, ratio metrics, geo-cluster similarity
- Evaluated using cross-validation and RMSE
- Saved deployable artifact with joblib
U.S. Flight Delay & Cancellation Analysis
- Analyzed 3 million U.S. flight records
- Identified delay and cancellation patterns by airline, route, season, and operational cause
- Engineered time-based features such as hour, weekday, and season
- Found late aircraft and carrier operations were major contributors to delay minutes
- Produced visual reports with actionable insights
Education
Arizona State University
Tempe, AZ
B.S. Business Data Analytics (STEM-Designated)
B.S. Artificial Intelligence (STEM-Designated)
Concurrent Degrees
Aug 2024 – Dec 2027
Get In Touch
Currently a Data Science Co-Op at Command Credit Corp — open to Summer 2027 AI engineering internships and opportunities.
Contact Information
Resume
Download my full resume to see a complete list of my skills, experiences, and academic achievements.
Download Resume