Software Engineer (Machine Learning).

Building and shipping consumer-facing products using AI-assisted development workflows and structured LLM evals.

1. CityBites — Restaurant Recommendation Site

   Full-stack recommendation site built end-to-end from a raw Google Maps Takeout CSV of 106 personal picks to a fully animated, searchable product.

   • Used AI-assisted workflows with structured evals to iterate and ship fast
   • 80 visitors within 24 hours of launch
2. WithSpaces — Productivity App

   Building an app for productivity in flow state; designed and shipped the waitlist experience.

Built ML systems for search relevance and ranking, collaborating with ML scientists and PMs to ship model improvements to production.

• Built Spark ML pipelines on Databricks for training and scoring search ranking models, including feature engineering to turn raw event data into model-ready ranking features
• Built second-pass reranking in the service layer with feature-serving endpoints and YAML-driven signal selection
• Reduced ML pipeline runtime from 15 hours to 2 hours (7.5× faster) by introducing Databricks Workflows to parallelise jobs
• Ran A/B experiments to measure model impact; integrated Travel Ads into the ranking backend with real-time placement logic
• Designed data services ingesting millions of events per day and optimised complex SQL queries powering ranking datasets

Built ML-based fraud detection systems and feature pipelines for real-time risk scoring at e-commerce scale.

• Built feature engineering pipelines transforming raw fraud signals into backend features consumed by risk models for real-time scoring
• Built ML-based post-processing for invoice fraud detection, applying model outputs to flag fraudulent invoices
• Designed and scaled APIs handling 250M+ orders and payments with low-latency fraud checks under heavy traffic
• Delivered a GDPR compliance project protecting user data across the risk platform

Expedia Hotels.com (personalised pricing APIs, Grafana monitoring pipelines), Profectus Solutions (0→1 pricing intelligence platform, led 8 engineers), Infosys (REST APIs with Kafka event streaming for Apple Concierge Analytics).