ChotuLab — AI Products for Real Life

About

The Engineer Behind ChotuLab

Sai Kumar Vurukonda

Senior ML Engineer

Fidelity Investments · 7 years experience

I build production AI systems at enterprise scale — fraud detection on graph neural networks over 50M+ node graphs, multimodal document intelligence processing 50K+ financial documents daily, recommendation engines fusing GraphSAGE with Transformer session modeling, and real-time audio ML pipelines handling 8M+ acoustic events daily. End-to-end ownership from distributed training through GPU-accelerated inference, with CI/CD model rollouts, drift monitoring, and regulatory compliance (FINRA, SOC2) across AWS and Kubernetes.

AWS Solutions Architect AWS Developer SAFe 5 Practitioner

MS Business Analytics — Trine University

MS Computer Science — Chicago State University

LinkedIn GitHub Email

🧠
Graph Neural Networks at Scale
Fraud detection on 50M+ node transactional graphs at Fidelity — improved precision-recall AUC by 28% over gradient-boosting baselines, reduced false-positive review volume by 40%.

📑
Multimodal Document Intelligence
Led FidScan — ColPali embeddings + GPT-4o Vision + OCR validation automating 50K+ KYC/AML documents daily. TensorRT quantization cut inference to sub-second, reducing manual review by 71%.

🎯
Recommendation & Experimentation
Next-Best-Action engine fusing GraphSAGE + Transformer session modeling. Designed A/B experiments for causal impact measurement and contextual bandit exploration for real-time action ranking.

🔊
Real-Time Audio ML Pipelines
At Shure — architected event-driven pipelines processing 8M+ acoustic events daily across enterprise audio platforms. CNN-based classification with 25% accuracy improvement in noisy environments.

⚙️
MLOps & Production Infrastructure
End-to-end ML platforms on AWS SageMaker + Kubernetes — MLflow tracking, feature stores, model registry, drift monitoring, CI/CD rollouts with rollback strategies across millions of transactions.

Our Mission

AI Should Solve Real Problems

Too many AI demos look impressive but never make it to production. At ChotuLab, every product is deployed, monitored, and built to the same standard as enterprise software — because that's the only kind that creates real business value.

We focus on domains where bad decisions cost money every single day — food waste, drug safety, financial risk, contract compliance. Places where AI can prevent loss, not just generate reports.

🎯

Production-First

Every project ships with CI/CD, Docker, monitoring, and a live URL. No Jupyter notebooks in demos.

📊

Measurable Impact

Each product is evaluated against real business metrics — MAPE, AUC, revenue impact, cost savings.

⚡

Full-Stack ML

Data pipeline → feature engineering → model training → API → dashboard. End-to-end ownership.

🔒

Secure & Lean

Secrets management, rate limiting, HTTPS, minimal cost. Enterprise quality without enterprise spend.

A Lab That Ships
AI That Works

The Engineer Behind ChotuLab

AI Systems Built for Real Impact

AI Should Solve Real Problems

A Lab That Ships AI That Works

The Engineer Behind ChotuLab

AI Systems Built for Real Impact

AI Should Solve Real Problems

A Lab That Ships
AI That Works