Shota Gen | Portfolio

Professional Experience

Pursuit

Software Engineer Summer 2025

Pre-RFP Intelligence Platform | Ann Arbor, Michigan

Technologies: Python, TypeScript, Docker, AWS (CloudWatch, EC2, RDS), LangSmith, Cursor

Architected a LLM evaluation platform using LangSmith and Docker, reducing benchmarking time by 90% (8 hours to 10 minutes) for 400+ experimental runs.
Engineered a 3-LLM chained analysis microservice (NestJS/TypeScript) that increased contact verification accuracy from 72% to 91% across 7.4M records.
Developed automated evaluation suites for crawler data extraction, contact verification, and document classification, accelerating prompt iteration cycles by 3x through granular error analysis.
Operationalized company-wide LLM traceability, implementing prompt versioning and configuration-driven model switching to track cost and latency for 100% of production runs.

Rakmy

Software Engineer Intern Summer 2024

Restraunt Data Platform | Shibuya, Tokyo

Technologies: Python, Anthropic API, AWS (RDS, CloudWatch, Lambda)

Built AI-assisted image data extraction with Anthropic API; reduced manual entry by 95%, achieved 90% extraction accuracy, and cut operational costs by an estimated $90,000 annually.
Shipped two data-wrangling APIs with custom validation to AWS RDS and deployed an image-extraction API on AWS Lambda; wrote implementation docs and test plans.
Designed a Slack-integrated chatbot for company-wide use that facilitates Claude prompting and thread summarization, enhancing communication efficiency across teams.

Rutilea

Project Manager Intern Summer 2022

AI Development Platform | Kyoto, Japan

Technologies: Python, Node.js, React

Spearheaded the end-to-end development of ’SmartBot,’ an AI-powered outsourcing service, leading a team of 2 to integrate a custom Slack chatbot with the Google Sheets API, which cut service delivery time by 40%.
Managed the project’s lifecycle using Agile methodologies and presented the business case to the CEO, positioning ’SmartBot’ as a viable new service line to drive company revenue.

Gnowbe

Front-End Development Intern Summer 2022

Microlearning Platform | Singapore

Technologies: Webflow, Canva, Loom, Adobe Illustrator

Developed interactive and stimulating UI components for the main page of their website using Webflow and Canva used by 2,000+ monthly visitors.
Created tutorial videos for the usage of Gnowbe’s microlearning platform using Loom for 50+ company clients and 1,000+ individual users.

Mars Flag

Full-stack Development Intern Summer 2021

Search Technology | Chuo, Tokyo

Technologies: Python, HTML, CSS, JS, SQLite, Flask

Developed website that recommends anime based on the user’s selected preferences utilizing full stack web development.

Technical Projects

StudyConnect

Winter 2026

Technologies: Swift, Python, PostgreSQL, ARKit, Google Cloud API, CoreLocation, Google Stitch, Figma, XCode

Developing a campus-centric iOS app for real-time peer coordination by mapping user presence within geofenced "Study Zones" and utilizing Ultra-Wideband (UWB) for precise indoor friend-discovery. As Product Manager, directing end-to-end development by leading team meetings, delegating technical tasks across the stack, and personally designing the UI/UX to ensure a seamless social mapping experience. Engineering an automated scheduling engine that leverages LLM tool calling and the Google Calendar FreeBusy API to analyze collective availability and propose optimal session times. Optimizing indoor navigation by fusing CoreLocation geofencing with a 60Hz sensor loop, incorporating ARKit, barometer data, and accelerometer inputs, to render stable 3D beacons through walls, overcoming GPS limitations in multi-floor buildings.

View Repository

This Portfolio Website

Winter 2026

Technologies: HTML, CSS, JS

Built a responsive personal portfolio website with collapsible sections for professional experience, technical projects, and school projects. Deployed with custom domain configuration on GitHub Pages and optimized for both desktop and mobile devices using modern CSS Grid and responsive design patterns.

View Repository

AutoVolume

MHacks 2025

Technologies: Swift, SwiftUI, AVFoundation, CoreAudio, macOS

Built a macOS menubar application that intelligently adjusts system output volume based on real-time ambient noise levels. The system continuously monitors microphone input using AVAudioEngine, calculates ambient dB levels via RMS analysis, and maintains a proportional volume-to-noise ratio to automatically adapt playback volume in dynamic environments. To prevent audio feedback loops, we established an initial noise-to-volume ratio at startup and applied consistent proportional scaling, avoiding reactive feedback where output volume increases cause the system to detect more noise. Integrated CoreAudio APIs to seamlessly control system volume while handling permissions and ensuring robust audio engine state management.

Main Use Case: So that we can hear our YouTube video while cleaning the dishes or vacuuming the floor.

View Repository

BookToManga

Winter 2026

Technologies: Python, Stable Diffusion, Meta SAM, LLMs

Automated the end-to-end conversion of long-form literary prose into serialized manga by engineering a multi-stage pipeline that transforms raw text into visually consistent comic panels. Developed a prompt engineering framework to instruct LLMs to generate granular storyboard descriptions, detailing character anatomy, foreground elements, and background environments, which served as inputs for custom Stable Diffusion models and LoRAs. Integrated Meta’s Segment Anything (SAM) and custom Python scripts to automate panel segmentation, dialogue bubble injection, and layout assembly.

AniRec (Pre-LLM)

Summer 2021

Technologies: Python, Flask, SQLite, HTML/CSS, JavaScript

Built a full-stack web application that recommends anime titles based on user preferences using genre filtering and search functionality. Processed a CSV dataset of 12,000+ anime entries into a SQLite database, then implemented dynamic genre-based filtering with partial string matching to surface relevant recommendations sorted by rating and popularity. Created an intuitive web interface allowing users to search by anime name or select multiple genres to discover top-rated titles. Overcame challenges in handling multi-genre anime entries and optimizing database queries for real-time search performance.

View Repository

ToDo List (Pre-LLM)

Winter 2017

Technologies: Swift, UIKit, Core Data, XCode

Developed an iOS to-do list application featuring date-based task filtering and persistent data storage using Core Data. Implemented swipe gesture navigation to allow users to browse tasks across different dates, with support for task creation, editing, and completion tracking via custom checkbox states. Designed a multi-attribute task system capturing goals, deadlines, consequences, and creation dates, leveraging delegation patterns for seamless view controller communication and data synchronization.

View Repository

Professional Experience

Technical Projects

School Projects