Rohan Sampath

Offline Translation: On-Device Image-to-Text Translator for iOS

An iOS app that translates text in images fully offline using on-device models—no internet required. Built in Swift, it supports three pipelines with the same UI/UX: (i) Apple-native components (Vision, Core ML, Translate), (ii) hybrid open-source models such as MLKit (OCR) and M2M-100 (Translation), and (iii) compact vision-language models for direct image-to-translation.

The real-world use case envisioned is travelers or users without an internet connection who need fast and private translation of text, such as road signs, menus, forms, or instructions.

Key research questions: Can a custom pipeline: (1) Translation Quality: Provide improved image-to-text translation accuracy for the languages currently supported by Apple? (2) Language Coverage: Provide working translation pipelines for languages not supported by Apple?

Keywords: CoreML, Apple Vision, Apple Language, Apple Translate, MLKit, OCR, M2M-100, miniGPT4, on-device translation, Swift, iOS, mobile AI, on-device models

(currently under review at the App Store)

ScriptForge: Agentic Screenwriting Assistant for Film & TV

A writer assistant that lets you co-author screenplays in multiple genres and formats with fine-grained control over collaboration. Uses a multi-agent architecture with a supervisor agent and multiple specialized worker agents. Built using LangGraph (soon moving to OpenAI Agents SDK), it orchestrates a multi-agent system where a human and LLM "Writer Agent" alternate turns in drafting, rewriting, and structuring scenes.

This approach preserves authorial voice and creative direction while drastically reducing the time to a complete screenplay. Includes tooling for test-time LLM selection, turn management protocols, and structured prompt libraries tailored for screenwriting.

Key differentiator: Configurable control over how often the LLM "Writer Agent" takes turns vs. the human writer — useful for both novice and expert screenwriters.

Keywords: LLM agents, LangGraph, LangSmith (observability), Model Context Protocol, multi-agent orchestration, OpenAI Agents SDK

(Repo currently private due to active commercial discussions.)

Head-to-Head Model Evaluation Harness

A lightweight head-to-head evaluation harness that lets you compare two LLMs—either off-the-shelf or fine-tuned—on a custom dataset. Designed for domain-specific testing (e.g., legal, finance, enterprise), and with curated test cases from non-technical users in mind. For example, a lawyer or financial analyst can define a small evaluation suite and compare model performance without writing code.

Built using vLLM - a high-throughput and memory-efficient inference engine for LLMs.

Key differentiator: Enables evaluation on custom datasets, not just public benchmarks — ideal for quick comparison and selection of models for specific domains or use cases.

Keywords: Hugging Face, vLLM, model evaluation, custom datasets, eval harness infrastructure

(Hugging Face Space - demo and proof of concept on a single A100 GPU.)

ThreeFortyThree Canada: Canadian Election Projections and Modelling

A seat-by-seat forecasting platform for Canadian federal elections. It was built in preparation for the 2025 Canadian federal election, held in April 2025. It combines regional and national polling with historical swing data, incumbent effects, demographic data, and multivariate modeling to simulate likely outcomes across all 343 electoral districts (hence the name). The projection engine uses Monte Carlo simulations to incorporate uncertainty and generate probabilistic seat counts and riding-level forecasts.

Users can interactively simulate elections by adjusting party vote shares and observing how seat-by-seat outcomes shift in response.

Key features: Uses a polling aggregation pipeline, combines multivariate distributions and Monte Carlo simulation to model multi-party dynamics. Read full methodology here

Keywords: React, Node.js, Python backend for Chat (see below, link), Multivariate distributions, Monte Carlo engine, Polling aggregation pipeline

Model Context Protocol (MCP) Server and Client: Elections Canada

A Model Context Protocol (MCP) implementation for structured, programmatic access to Canadian federal election data—built to support agentic reasoning in a chat interface.

Server: The MCP Server exposes detailed election results across Canada's 343 electoral districts. While the server was specifically tested for use with Claude Desktop, the server can power any MCP-compatible client.

Client: A multi-agent MCP client built with LangGraph, powered by the above MCP Server. See ThreeFortyThree Chat.

Key feature: One of the first open MCP implementations designed for political (election results) data.

Keywords: MCP, Model Context Protocol, Canadian elections, Claude Desktop, multi-agent architecture, LangGraph

Server: PyPI Client: