I design and ship reliable systems: data pipelines, streaming services, and APIs that turn messy inputs into clean, fast, observable products. 3+ years in the trenches, currently finishing a Master’s in Software Engineering @ UHCL. I care about smart abstractions, tests that matter, and shipping code that’s easy to run at 3 a.m.

• 🔭 Building: event-driven backends, real-time analytics, resilient pipelines
• 🧪 Loop: model → automate → deploy → measure
• ☁️ Clouds: AWS (S3, EMR, Glue, Lambda, Redshift), GCP
• 📍 Houston-ish • Remote-friendly

Languages: Python, SQL, Java, JavaScript, Typescript
Backend & APIs: FastAPI, Node.js, REST
Streaming & Compute: Kafka, Apache Spark, AWS Glue/EMR/Lambda
Data Stores: PostgreSQL, Redshift, DynamoDB, MongoDB, Firebase
Modeling & Orchestration: dbt, Apache Airflow
DevOps: Docker, GitHub Actions, Prometheus, Grafana
Frontend (enough to ship): React.js
Testing: PyTest, Postman, smoke tests, contract tests

Repo: fraud_detection
When: May 2025

Automated anomaly detection pipeline using Isolation Forest (PyOD) for financial transactions — no labels needed.

• Real-time scoring with configurable contamination rate
• Visual analytics: normal vs. anomalous distributions
• Clean notebooks and reproducible environment

Stack: Python, Pandas, NumPy, Matplotlib, Seaborn, PyOD, Jupyter

Production Q&A over unstructured PDFs with low-latency retrieval, stable ingestion, and observability.

• Sub-200ms answers, doc auto-ingest, dashboards and metrics
  Stack: FastAPI, ChromaDB, Prometheus, Grafana

ETL → normalized DB → executive dashboards. Computes PER, TS%, and Four Factors with rate-limits and retries.

Stack: Python, SQL, Power BI

Full-stack patient management with RBAC/JWT and real-time alerts. Reduced scheduling time by ~45%.

Stack: React, Node, Postgres, Socket.io, AWS SNS

Automations across Google Sheets/CRM. Standardized inputs led to ~30% fewer errors and ~40% faster updates.

React + MUI front ends integrating Stripe/Razorpay and Google Maps. Reporting scripts turned hours into minutes.

• Event-first modeling: design streams and schemas before code
• Contracts & tests: JSON schema + smoke + contract tests
• Guardrails: idempotent jobs, checkpoints, safe backfills
• Observability: metrics over vibes; useful dashboards and alerts

• Event-driven system design (sagas, outbox, backpressure)
• Cost-aware modeling on Redshift / column stores
• Better DX for dbt + Airflow in mono-repos

• 🔗 LinkedIn: https://www.linkedin.com/in/krunalchauhan/
• 💻 GitHub: https://github.com/krunal96369
• ✉️ Email: krunal96369 (at) gmail (dot) com