Aayush Tiwari

Helped develop a sanctions-screening proof of concept into a configurable platform, covering processing pipelines, system design, and operational tools.

• Participated in system-design proposals with engineering and project stakeholders, translating operational requirements into architecture and workflow.
• Built and iteratively evolved a Python, FastAPI, React, and TypeScript platform capable of processing files with 80M+ records across DEV, UAT, and pre-production environments.
• Implemented configurable ingestion, validation, normalisation, clustering, candidate generation, rules, and decision-merging stages.
• Built run monitoring, warnings, timelines, and paginated outputs for the operational dashboard.
• Added run-scoped isolation, bearer authentication, RBAC, audit logging, and Pytest-based testing.

Refactored a Node.js backend for an AI counselling product, separating inference from session state and simplifying a four-stage conversational workflow.

• Reduced inter-service coupling by approximately 40% and code redundancy by approximately 30%.
• Implemented keyword-triggered crisis routing with a routing decision under 100 ms.
• Built a profile-aware returning-user workflow that removed redundant onboarding for approximately 60% of repeat sessions.

Brought operational data into a SQL warehouse and built Power BI reporting for revenue, churn, and sales-funnel analysis.

• Integrated 5+ source systems into a normalized, fact-and-dimension data warehouse.
• Delivered 3+ Power BI dashboards with 15+ custom DAX measures for revenue, churn, and funnel analysis.
• Reduced ad-hoc reporting turnaround by approximately 35% and improved refresh time by approximately 25% through query and model optimization.

A research system that compares four specialised AI agents with simpler rules for ranking Indian mutual funds. Includes reproducible evaluation and an inspection dashboard.

Python / Ollama / Qwen2.5-7B / Gemini / Pandas

A local incident-investigation platform that brings metrics, logs, traces, service topology, and changes together to explain likely causes with supporting evidence.

Python / FastAPI / Next.js / PostgreSQL / pgvector

An academic collaboration using Kafka and Spark to score transaction streams. A simulated local benchmark tests whether fewer features can reduce inference cost.

Apache Kafka / PySpark / Spark Structured Streaming / Spark MLlib / NumPy