Looking for a job

Tejasv Agarwal

@tejasvagarwal

Message

I build machine learning pipelines and full-stack data platforms for real-world sensor and AI applications.

United States

Message

What I'm looking for

I'm looking to build practical machine learning, data, and full-stack systems where I can improve model performance, ship reliable products, and continue growing through challenging real-world work.

I'm building machine learning pipelines at Pennsylvania State University that turn machining sensor data into CWT/FFT scalograms for CNN-based process classification.

I benchmarked ResNet-18/50, EfficientNet-B0, and ConvNeXt-Tiny across 128+ configurations in PyTorch, reaching 98% spindle-speed classification accuracy. I also improved feed-rate classification from 54.2% to 87.5% through condition-residual preprocessing and leave-one-side-out validation.

At CAIS Lab, I built an end-to-end Python data platform for 30K+ sensor records, combining SQLite, preprocessing pipelines, automated validation, FastAPI services, and an interactive JavaScript dashboard. My reusable modeling workflows improved time-series prediction MAE by 74.3% over a mean baseline.

I've also built AlphaNexus, a full-stack financial backtesting engine, and JobFit AI, a recruitment matching platform with explainable fit scores. Earlier, I developed OCR, NLP, and RAG workflows at Outamation, improving document-retrieval accuracy from 65% to 92%.

Experience

Work history, roles, and key accomplishments

Current

Undergraduate Research Assistant

Current

Pennsylvania State University

Apr 2026 - Present (6 months)

Developed an ML pipeline converting machining sensor data into CWT/FFT scalograms for CNN-based process classification. Benchmarked 4 CNN architectures and improved feed-rate classification accuracy from 54.2% to 87.5%.

Python PyTorch CNN ResNet EfficientNet ConvNext SMOTE Signal Processing CWTS FFT

Data & Backend Engineering Intern

CAIS Lab

Jun 2026 - Jul 2026 (1 month)

Built an end-to-end Python data platform for 30K+ sensor records, integrating SQLite storage, preprocessing pipelines, FastAPI services, and an interactive JavaScript dashboard. Engineered validation and modeling workflows that improved time-series prediction MAE by 74.3%.

Python SQLite fastAPI JavaScript Data Engineering Time Series Analysis Cross Validation

Undergraduate Teaching Assistant

Pennsylvania State University

Aug 2025 - May 2026 (9 months)

Elevated average quiz scores by 15-20% for 50+ Calculus I students by leading weekly interactive recitations and targeted office hours. Drove a 30% increase in student participation through collaborative group activities.

Teaching Mentorship Calculus Communication Skills Group Facilitation

Artificial Intelligence Externship

Outamation

May 2025 - Jul 2025 (2 months)

Designed an OCR + NLP pipeline for mortgage document automation using PyMuPDF and Tesseract. Built a RAG system with LlamaIndex and implemented hybrid retrieval improving accuracy from 65% to 92%.