nv://agent · online · open to new-grad AI/ML & SWE roles

Building production-grade intelligent systems.

I'm Nil Vaghela — a Computer Science graduate from Georgia State University. I design and ship ML pipelines, RAG applications, backend services, and quantitative research, with internships at DXC Technology and Northern Trust.

View Projects Download Résumé Contact
0
Industry internships
0
Records pipelined
0
Bars in research
0
Published paper
Scroll

01 — Selected Work

Things I've designed,
built, and shipped.

Production-style systems — not class assignments.

01 Live

Prepare-Up — RAG-Powered Study Platform

Full-stack AI study platform that turns uploaded documents into grounded chat, flashcards, study guides, and podcast-style learning content. Built on a retrieval-augmented pipeline (FAISS + LangChain + OpenAI) with a FastAPI backend, PostgreSQL persistence, and a React frontend deployed on AWS.

FastAPIPostgreSQLOpenAI APILangChainFAISSReactAWS
GitHub
02 Research

Greeks-Aware Options Signal Research — Quant ML Research (SSRN)

ML framework for NIFTY 50 intraday options trading using engineered Greeks, volatility regimes, walk-forward validation, Black-Scholes P&L simulation, GARCH volatility estimation, and ensemble models (XGBoost + LightGBM) tuned with Optuna. Trained and evaluated on 122K+ intraday bars with leakage-aware, regime-split validation.

PythonXGBoostLightGBMOptunaGARCHBlack-ScholesQuant ML
Read on SSRN
03 In Build

ClosetEye — AI Wardrobe Recommender

Mobile-first AI wardrobe system: computer vision for clothing classification, visual similarity search for 'matches what I already own', outfit recommendations, and FastAPI-backed inference. React Native frontend, PyTorch CNNs serving inference.

PyTorchCNNsFastAPIReact NativeComputer Vision
GitHub
04 Shipped

Data Automation & Backend Systems — ETL · APIs · Analytics

Collection of ETL pipelines, REST APIs, SQL reconciliation workflows, and analytics dashboards built across internships and independent work — operating against 500K+ record datasets and production-style infrastructure on AWS and Snowflake.

PythonSQLSnowflakeAWSDockerPower BIETL
GitHub
05 Research

Wearable Diabetes Risk Screening — Clinical + Wearable ML Pipeline

Two-layer data-mining pipeline for diabetes risk screening (triage, not diagnosis): a clinically grounded baseline classifier trained on the Pima Indians dataset, plus a wearable-to-clinical feature-translation model that enables wearable-only risk inference. Built with stratified 60/20/20 splits, leakage-aware validation, decision-threshold tuning on a held-out set, and final reporting via ROC-AUC, precision/recall, F1, confusion matrices, and calibration curves.

Pythonscikit-learnPandasNumPyJupyterModel Calibration
GitHub
06 Shipped

Monte Carlo Simulator — Interactive Market-Risk App

Interactive Streamlit application that runs Monte Carlo simulations on historical market data — pulled via yfinance and cached locally in SQLite — to project possible future price paths and quantify downside risk. Packaged with a dev-container for reproducible, one-click runs.

PythonStreamlitNumPyPandasyfinanceMonte Carlo
GitHub

02 — Experience

Where I shipped
real systems.

Internships against real data and production-style infrastructure.

  1. AI/ML & Software Engineering Intern

    DXC Technology

    May 2025 – Jul 2025 Atlanta, GA
    • Built Python and SQL-based ML/data pipelines operating across 500K+ enterprise records.
    • Developed ETL workflows connecting Snowflake with AWS S3 / EC2.
    • Improved data consistency and supported KPI forecasting workflows used by ops.
    • Exposed model outputs through APIs and dashboards for operational visibility.
    PythonSQLSnowflakeAWSETL

  2. Data Science / Automation Intern

    Northern Trust

    May 2024 – Jul 2024 Chicago, IL
    • Built transaction-data pipelines and fraud-risk modeling workflows.
    • Engineered time-series features — rolling aggregates, lag features, velocity indicators.
    • Automated SQL reconciliation workflows and reduced manual reporting effort.
    • Evaluated classification models using ROC-AUC, F1-score, and precision/recall.
    PythonSQLTime-SeriesScikit-learnAutomation

03 — Toolbox

What I reach for
when I'm shipping.

01Languages

PythonSQLJavaScriptTypeScriptC++

02AI / ML

Scikit-learnXGBoostLightGBMPyTorchCNNsFeature EngineeringModel Evaluation

03GenAI / RAG

OpenAI APILangChainFAISSEmbeddingsPrompt EngineeringRAG Pipelines

04Backend / Data

FastAPIREST APIsPostgreSQLSnowflakeETLData Modeling

05Cloud / DevOps

AWS S3AWS EC2DockerGitHub ActionsCI/CDJenkins

06Analytics / Tools

PandasNumPySciPyMatplotlibPower BIGit

04 — How I build

Designed for
evaluation, not demos.

I architect intelligent systems from first principles — retrieval pipelines, ML inference services, and quant-grade evaluation harnesses that hold up outside a notebook.

  1. 01

    Honest evaluation first.

    Walk-forward validation, leakage audits, and real baselines before any model claims.

  2. 02

    Compose small services.

    FastAPI + queues + workers. Each model is a single-responsibility endpoint behind a typed contract.

  3. 03

    Ground retrieval, don't fake it.

    FAISS + reranking + citation-aware prompts so LLM output is inspectable, not vibes.

05 — Research

Quant ML, published.

SSRN Quant ML Options Ensemble Learning

Greeks-Aware Intraday Options Signal Generation on NSE Nifty 50 Using Ensemble ML

Applied ML research using Greeks-aware feature engineering, Black-Scholes P&L simulation, GARCH volatility estimation, ensemble models, Optuna hyperparameter tuning, and walk-forward validation across 122K intraday bars. Designed for honest evaluation — no look-ahead bias, regime-aware splits, and execution-cost-aware backtests.

Read on SSRN Methodology

06 — About

Systems that are
sound, usable,
and measurable.

I'm a Computer Science graduate from Georgia State University focused on building practical AI/ML and backend systems. My work spans RAG applications, ML pipelines, data automation, and quantitative research.

I care about systems that are technically sound and honest — careful attention to data leakage, real baselines, and infrastructure that holds up outside a notebook. Through internships at DXC Technology and Northern Trust, I've worked on enterprise-scale data pipelines, fraud-risk modeling, ETL between Snowflake and AWS, and production-style backend services.

07 — Education

Foundations.

B.S. Computer Science

Georgia State University

Graduating May 2026 · Atlanta, GA

Relevant coursework edit me

Data Structures & AlgorithmsMachine LearningDatabase SystemsOperating SystemsSoftware EngineeringLinear Algebra

08 — Get in touch

Let's build something intelligent.

Open to AI/ML, backend, data, and quant-focused new-grad roles. Drop me a note below, or reach out directly.

or reach me directly

Email Me LinkedIn GitHub
nilrajsinh060@gmail.com