Selected work from my independent projects and enterprise engagements:
Developed a revolutionary real-time anomaly detection platform that prevents fraud, quality issues, and security breaches across multiple industries. Built scalable streaming data processing architecture using Apache Kafka and TensorFlow, enabling instant detection of anomalies in financial transactions, manufacturing processes, and healthcare monitoring. The system processes millions of data points per second with sub-second latency, preventing millions in potential losses.
Tech Stack: Python, Apache Kafka, TensorFlow, Redis, Docker, Machine Learning
Impact: Prevents fraud losses, improves quality control, reduces operational costs
Business Value: Scalable across finance, manufacturing, healthcare, energy, telecom, retail, and government sectors
Built a comprehensive privacy-aware data analytics platform with enterprise-grade security. Features differential privacy for customer data analysis, automatic PII detection and masking, role-based access controls, and full GDPR/CCPA compliance. Enables powerful customer behavior analysis, trend detection, and real-time dashboards — all without exposing personally identifiable information.
Tech Stack: Python, FastAPI, PostgreSQL, React.js, Chart.js, Docker
Impact: Privacy-preserving analytics with differential privacy, PII masking, and audit logging
Business Value: Enterprise-ready compliance (GDPR/CCPA) with full data lineage and privacy budget tracking
Batch pipeline that ingests market data after every close, validates it at ingestion, computes point-in-time features and writes ranked output for a stock analytics platform. An LLM layer turns the structured output into plain-language analysis.
Tech Stack: Python, Pandas, NumPy, Anthropic API, pytest
Impact: Fully automated nightly runs with validation at ingestion
Backtesting framework with strict point-in-time controls that prevent lookahead bias, behind a modular data-source interface so providers can change without touching downstream logic.
Tech Stack: Python, Pandas, pytest
Impact: Leak-free historical evaluation and swappable data providers
Postgres data model and storage layer for a subscription platform, with signed-URL access control for gated content and payments in multiple currencies.
Tech Stack: PostgreSQL (Supabase), Cloudflare R2
Impact: Users in 8 countries
Agentic AI system that runs on its own schedule: it scans live markets every few minutes, scores opportunities with predictive models, sizes positions, places and exits orders through an exchange API, and reconciles fills and P&L. Every decision is logged with its reasoning for audit.
Tech Stack: TypeScript, Cloudflare Workers, Durable Objects, D1, Python
Impact: Fully autonomous decisions and execution, with a full audit trail
End-to-end Azure data pipelines with PySpark, Databricks and dbt processing millions of manufacturing records a day, plus TB-scale ingestion and transformation frameworks that became the backbone of enterprise reporting.
Tech Stack: Azure, Databricks, PySpark, dbt, Azure DevOps
Impact: 30% lower data latency; dashboards used daily by 50+ managers
Automated PySpark pipelines on Databricks and a layered medallion data model on Delta Lake for drug product data, with rule-driven validation on stability testing data.
Tech Stack: Databricks, Delta Lake, PySpark, dbt, ADF
Impact: 40% faster processing; zero defects at launch across 200+ validated fields
Led a five-person team to build a Snowflake analytics platform consolidating 12 siloed sources into one enterprise data model, and redesigned global KPI reporting across 4 business units.
Tech Stack: Snowflake, Azure, Python, SQL, Tableau, Power BI
Impact: Single source of truth; 80% faster executive reporting