Saahil MehtaAssociate Director, Data & Insights

Community

Mentoring, a museum archive, and a developer programme.

  1. 2025 - Present

    MERL Visitor Insight Engine

    Museum of English Rural Life · Volunteer

    Nine years of handwritten visitor feedback sitting in physical comment books. Cursive, children's scrawls, notes in Arabic, Mandarin, Turkish, German. References to philosophy, politics, internet memes. Material traditional OCR can't read.

    The Pipeline

    A two-tiered agentic system. The Visual Reasoning Engine processes each scanned page with language-agnostic handwriting recognition and contextual understanding. In one entry, it recognised a string of Chinese characters as Zhang Zai's 11th-century Four Sentences (a Neo-Confucian philosophical text) and reasoned about why a visitor might inscribe this in a museum about English rural life.

    The Deep Research Agent synthesises thousands of entries and surfaces patterns that manual review would miss: a single viral tweet transforming visitor demographics, political events in Turkey appearing in a Reading guest book around the election, visitors from over 75 countries connecting their own agricultural heritage to English farming history.

    What This Proves

    The system reads handwritten text and understands the context behind it.

    When it encountered Turkish handwriting mentioning "Kılıçdaroğlu", it recognised this as a reference to the 2023 presidential election and understood why a visitor expressed political sentiment at that moment. It parsed internet slang like "L + ratio" as meme-culture discourse and analysed the social dynamics at play. When a visitor from St Helena (population 4,500, remote South Atlantic) wrote about unexpected connections to home, it grasped the geographic and cultural weight of that moment.

    Contextual intelligence, working across languages, cultures, and time periods.

    Outcome. Delivered a strategic analysis to museum leadership covering sentiment trends, demographic insights, and visitor journey mapping against the Generic Learning Outcomes framework. Work that would have taken months of manual effort, completed in days. The museum now has evidence-based insight into nearly a decade of visitor experience, unlocked from a previously unreadable archive.

    Visual ReasoningMultimodal LLMsContextual IntelligenceDeep Research Agents
  2. 2026 - Present

    Data Science Festival Mentorship

    Data Science Festival · Mentor

    Spring 2026 Cohort: I mentor people moving into data science, engineering, and AI roles via the Data Science Festival, the UK's largest free data science community. Topics I cover most: profile review, career progression, the technical bar at each level, and the actual interview pipeline.

    MentoringData ScienceCommunity
  3. 2026

    Google Developer Specialist

    Google · Invite-only Programme

    Selected for Google's invite-only Developer Programme and recognised as a Google Developer Specialist, contributing to the Google Developer community through technical content on Medium.

    Google DeveloperTechnical WritingCommunity

Let's talk.

About building AI systems people actually use, leading data teams, or your data and cloud estate.

LinkedIn Book a call

Also available asresume.pdfresume.jsoncv.mdllms.txtstatus.json