Hello, I'm Raaj Patel
I build AI thatworks in reality.
AI/ML Engineer & Research Assistant. Building end-to-end ML systems across time-series anomaly detection, LLM/RAG applications, and computer vision.
Raaj Patel
AI Researcher · Engineer
Move your cursor to explore
IEEE Published
SETCOM 2025
DER Anomaly Detection
Transformer F1 0.8182
Core Stack
Python + PyTorch
LLM Systems
RAG + Text-to-SQL
Inside the work
01 / APPROACHEngineer · Researcher · Builder
Models matter. Systems make them useful.
I'm an M.S. Computer Science student and AI/ML Research Assistant focused on applied AI systems, not just models. My work spans time-series anomaly detection for cyber-physical systems, LLM systems (RAG, Text-to-SQL, prompt engineering), and computer vision. I care about grounded AI where LLMs are tied to real data, reproducibility in research, and simulation-backed experimentation. Before grad school, I earned my B.Tech in ICT from PDEU (9.68/10 CGPA) and gained industry experience at Einfochips (Arrow Electronics) and Navarang Engineering Works. I have a published IEEE paper on forest fire severity detection.
Featured
3Focus areas that define my research and engineering lens.
Cyber-Physical Anomaly Detection
Simulation-grounded DER research benchmarking 6 detector families; Transformer @ 60s reaches F1 0.8182.
SQL-Grounded RAG
LLM explanation tied directly to SQL retrieval so numeric answers stay auditable, not generated.
IEEE Publication
Forest fire severity detection pipeline published at SETCOM 2025 (IEEE Xplore).
Selected Work
3Research and engineering projects across time-series anomaly detection, LLM systems, and computer vision.
DERGuardian — Cyber-Physical Anomaly Detection
Canonical benchmark: Transformer @ 60s window, F1 = 0.8182, across 6 detector families and 4 time windows.
- ●Canonical benchmark: Transformer @ 60s window, F1 = 0.8182
- ●6 detector families benchmarked: threshold, Isolation Forest, autoencoder, GRU, LSTM, Transformer
- ●IEEE 13-bus and 123-bus OpenDSS simulations with PV/BESS assets
AI Expense Tracker with RAG-Enhanced Financial Reasoning
Demonstrates database systems + AI integration with production-style thinking, not toy ML.
- ●Natural language -> SQL -> retrieved context -> grounded explanation pipeline
- ●All numeric values sourced from SQLite records, not generated by the LLM — reducing numerical hallucination risk
- ●Single integration point across 5 LLM backends: Ollama, OpenAI, Groq, HuggingFace, vLLM
Forest Fire Severity Detection (IEEE Publication)
Published: P. K. Barik, J. Suthar, and R. Patel, "Forest Fire Severity Detection using AI," SETCOM 2025, IEEE. DOI: 10.1109/SETCOM64758.2025.10932627.
- ●Published: SETCOM 2025, IEEE. DOI 10.1109/SETCOM64758.2025.10932627
- ●End-to-end CV pipeline: preprocessing, augmentation, training, evaluation
- ●Research-grade experimentation with iterative hyperparameter tuning
Experience
3Research and industry roles focused on AI systems, simulation, and engineering delivery.
Graduate Research Assistant — AI and Cyber-Physical Systems · Purdue University Northwest
Jul 2025 - PresentAdvisor: Prof. Shafkat Islam
- ●Design time-series ML detection pipelines for cyberattacks in Distributed Energy Resource (DER) systems using high-frequency electrical telemetry.
- ●Build OpenDSS simulations of IEEE 13-bus and 123-bus networks with integrated PV and battery storage (BESS) assets for controlled anomaly studies.
- ●Benchmark 6 anomaly detectors across 5s, 10s, 60s, and 300s windows; a Transformer at 60s achieved benchmark F1 = 0.8182 on the canonical evaluation.
- ●Generate and validate attack scenarios covering false data injection, spoofing, and command manipulation, with reproducible preprocessing and evaluation pipelines.
Software Developer (Contract) · Navarang Engineering Works
Jul 2024 - Jul 2025- ●Built and deployed the company website end to end and introduced digital tooling that reduced manual record-keeping.
Engineering Intern · Einfochips (Arrow Electronics)
Jan 2024 - May 2024- ●Supported AI and embedded workflows through Linux automation scripting, debugging, and structured test execution.
Education
2Graduate and undergraduate training with a focus on AI/ML systems.
M.S. in Computer Science
Expected May 2027Purdue University Northwest, Hammond, IN
GPA: 3.7 / 4.0B.Tech in Information and Communication Technology
May 2024Pandit Deendayal Energy University (PDEU), Gandhinagar, India
GPA: 9.68 / 10 CGPASkills
5A balanced stack across AI/ML, LLM systems, and production-ready engineering.
Languages
Machine Learning
LLMs & Generative AI
Frameworks & Libraries
Data & Systems
Publications
1Peer-reviewed work that demonstrates research rigor and delivery.
2025 International Conference on Sustainable Energy Technologies and Computational Intelligence (SETCOM), IEEE · 2025
Forest Fire Severity Detection using AI
P. K. Barik, J. Suthar, and R. Patel
DOI: 10.1109/SETCOM64758.2025.10932627
View on IEEE XploreNow
What I am focused on this semester.
- ●Building the DERGuardian detector benchmark and held-out synthetic evaluation
- ●Improving grounded RAG evaluation for the AI Expense Tracker
- ●Outside of research: reading, spending time with friends, and staying curious
Contact
Open to AI/ML engineering roles, research collaborations, and high-impact projects.