Chicago · Open to meaningful AI work
AI engineer · systems builder · problem solver
I build AI that
earns trust.
Enterprise RAG, agentic workflows, document intelligence, and production ML—designed for real users, real constraints, and measurable outcomes.
AI You are chatting with an AI-generated digital version of Parth, not the live human.
Standing byMove your pointer—Parth AI follows your attention.
01AI SYSTEMS
The point of view
Research is only useful
when it survives reality.
I'm Parth Oza, an AI Engineer based in Chicago. I translate complex machine learning ideas into dependable systems—grounded answers, observable pipelines, responsible automation, and interfaces people can actually use.
My work sits where ambitious AI meets enterprise responsibility: financial data, audit requirements, deployment constraints, and the humans who depend on the result.
Parth Oza AI Engineer
Selected systems
Work with consequences.
Built across retrieval, orchestration, ML pipelines, cloud deployment, and human-in-the-loop decisions.
01
Enterprise AI · Northern Trust
Conversational RAG for financial workflows
Designed and deployed a RAG-based conversational system with LangChain and Azure OpenAI, paired with evaluation for hallucination detection and response quality.
- Azure OpenAI
- LangChain
- RAG evaluation
- AKS
35–40%estimated reduction in manual handling
02
Document intelligence
OCR + LLM document automation
Extracted, classified, and routed structured information from unstructured financial documents with audit-conscious processing.
- OCR
- LLMs
- Python
- Azure
20+ hrsestimated weekly manual review eliminated
03
Applied machine learning
Candidate scoring system
Built an end-to-end XGBoost pipeline and FastAPI scoring service for faster, more consistent candidate review.
- XGBoost
- FastAPI
- Pandas
- Scikit-learn
82%held-out prediction accuracy across 500+ profiles
04
Deep learning · Time series
Market trend forecasting
Trained a PyTorch LSTM using five years of OHLCV history and 15+ engineered market indicators, with a visual backtesting workflow.
- PyTorch
- LSTM
- yFinance
- Matplotlib
~74%directional accuracy in the documented test period
Experience
From prototype
to production.
Since 2022, building applied AI across enterprise finance, retail, document workflows, and cloud ML platforms.
Jul 2024 — Present
AI Engineer
Northern Trust · Chicago
Conversational RAG, large-scale financial data pipelines, AI document processing, containerized inference on AKS, prompt evaluation, and enterprise security collaboration.
Jan — Jun 2025
Chapter President
NSLS · Roosevelt University
Led initiatives for 100+ members, organized leadership and technical programs, and connected emerging AI topics with professional development.
Nov 2022 — Nov 2023
ML / AI Engineering
ValueLabs · India
FAISS-based product matching, LLM document automation, cloud data pipelines, explainability interfaces, and client delivery across concurrent projects.
Technical practice
The stack is a means.
The system is the work.
LLM systems
RAG architecture, LangChain, LangGraph, LlamaIndex, prompt engineering, evaluation, vector search, fine-tuning.
Agentic workflows
Stateful orchestration, tool use, multi-agent patterns, memory, human review, and reliable failure paths.
Cloud & MLOps
Azure, AWS, FastAPI, Docker, Kubernetes, Airflow, CI/CD, MLflow, monitoring, and production inference.
Applied ML
PyTorch, TensorFlow, XGBoost, NLP, computer vision, time series, feature engineering, and explainability.
Python · LangGraph · Azure OpenAI · PyTorch · FastAPI · FAISS · AWS · Docker · Kubernetes · PostgreSQL · MLflow · SHAP · Python · LangGraph · Azure OpenAI · PyTorch · FastAPI · FAISS · AWS · Docker · Kubernetes · PostgreSQL · MLflow · SHAP ·
Education
Computer science,
data, and leadership.
M.S. Computer ScienceRoosevelt University · Chicago · Dec 2025
B.Tech Computer EngineeringS.P.B. Patel Engineering College · Gujarat Technological University · 2023
MLOps · Generative AI · IBM AI Developer · Google Data Analytics
The person behind the AI
Original moments.
Real perspective.
These photographs come directly from Parth's current portfolio and sit alongside the generated AI scenes as a clear connection to the real person.
Let's build something useful
Bring me the difficult problem.
Based in Chicago. Open to AI engineering roles, product collaborations, and conversations about applied LLM systems.