Piyush Satti

I'm a Python developer at Citylogix, where I started in February on an Airflow pipeline that processes road-survey imagery. I containerised its image-processing and GIS steps with Docker and owned the integration of a new data-collection vehicle. I also researched and implemented a panoramic reprojection method that turns 360° imagery into planar views, so field vehicles can carry fewer cameras, and researched and validated an inference method that improved detection quality; it was adopted in production. From August most of my time went into a platform that moves around 12 TB of field survey data a day into long-term storage, which I built end to end from design to production. These days I build training-data tooling with the AI team: a database, a record format and the command-line tools around them. I designed, directed and reviewed AI-written code for all of it, deciding what gets built and how it fits together while the agents do the typing.

Remote. Owned end-to-end delivery for client-facing software systems, from ambiguous requirements through design, implementation, iteration, and handoff under fixed deadlines.

• Built a vision-based game QA automation tool using OpenCV for on-screen detection and PyAutoGUI for input control to automate regression flows
• Shipped an AI movie recommendation platform with an ML-based ranking pipeline, a React web UI, and a Node.js backend exposing APIs for search and recommendations
• Led small project teams on deadline-critical engagements to meet client delivery targets

Supported Object-Oriented Programming II by debugging student code, explaining OOP concepts, and helping students implement course projects.

• Programmer on Duty for the course, providing hands-on debugging and implementation support
• Led weekly sessions for ~20 students and delivered a revision lecture for ~40 students covering core Java/OOP topics

Conducted applied research in algorithm-based image restoration and signal processing, translating papers into reproducible implementations and evaluations.

• Published 6 peer-reviewed papers and achieved state-of-the-art results in image restoration
• Implemented 40+ research techniques and evaluated improvements using signal-to-noise ratio metrics
• Contributed to a peer-reviewed publication (IEEE SPL)