0%

Rahulraj P V

Rahulraj P V

Chapter II · A question

“One day we asked ourselves:
how do we see what we see?”

Chapter III · Cinema

“Films became my answer, and my passion.
I wanted to make them. But first…”

Chapter IV · Physics

“…I was studying physics. Around then,
we started asking how people understand films.”

Chapter IV

“With friends, we ran experiments
and analysed how people watch.”

Every result gave us a rush of dopamine.

Chapter V

“To understand the patterns, I joined a PG Diploma in Data Science
and Analytics at Kannur University, Mangattuparamba campus.”

My first skills in AI and data science. A second chance in life.

Chapter VI

“And later, I joined Digital University Kerala for an MTech.
Guided by Professor Asharaf S, I changed a lot.”

New skills in AI and data science, and many late nights.

Chapter VII

“But I felt I had to learn more,
to truly understand human perception.”

So I wrote to Professor Tim J. Smith at the Creative Computing Institute, University of the Arts London.

Chapter VIII

“Then the email came: a place at UAL,
with its highest scholarship, £25,000.”

An MRes in Creative Computing, with Tim as my supervisor. I read it three times. The hard work had paid off.

Chapter I · Home

“I grew up in Kannur,
in the north of Kerala, India.”

Chapter VIII · Me

This is my story. My name is

Rahulraj P V

Latest projectpitute.A network that finds researchers by the work they do. Log in to pitute.org
A wooden 1970s television set
on
vol
Gradient, an AI newsroom that puts itself to air. Upper knob is power, lower is volume.

Rahulraj P V

I study how people watch films, and build machines that learn to watch with them.

Researcher at the Creative Computing Institute, University of the Arts London, with Professor Tim J. Smith. I build pitute, Gradient, Madhuri and ShadowStage.

Researcher and creative technologistKerala to London

Selected projects

Six folders, each one a thing I built

01pituteA network that finds researchers by the work they do. 02GradientA newsroom of AI agents that puts itself to air. 03MadhuriA campus radio station, run by AI. 04ShadowStageTholpavakkoothu, Kerala’s shadow puppetry, in code. 05DinoSynthesisA dinosaur’s call, made by a model. 06windmill.aiTalks into slides: code, timers and all.
pitute.org, the landing page: the map of open research opportunities, sign-up, what pitute does, and what early testers say.

pitute.org, the landing page as it is today. Scroll the screen.

What pitute does

One place to find research opportunities, keep up with your field and apply on time.

Before the show

What Tholpavakkoothu is

Tholpavakkoothu takes its name from thol, leather, pava, puppet, and koothu, play. It is the shadow puppet theatre of the Palakkad, Thrissur and Malappuram districts of Kerala, and it is played as an offering to the goddess Bhadrakali. The story goes that she was away fighting the demon Darika when Rama killed Ravana, so the war is played again for her, with her idol set on a pedestal facing the screen.

It is performed in a temple theatre, the koothumadam, a long stage with a white cloth stretched across it. Behind the cloth, 21 lamps of halved coconut shell or clay sit evenly along a wooden beam, the vilakku madam. The puppets, cut from goat skin (once deer skin), painted with vegetable dyes and some four feet tall, are held flat against the cloth on two sticks, one for the body and a thinner one for a limb, so each throws a sharp shadow and its punched holes turn into points of light.

The text is the Kamba Ramayana, sung and argued in a mix of Tamil, Malayalam and Sanskrit by the pulavar, the lead puppeteer and a scholar, who learns more than 3,000 verses over six to ten years of training. A full telling uses 180 to 200 puppets and up to 40 artists, runs 21 nights from dusk to dawn, and plays between January and May to the chenda, maddalam, ezhupara, ilathalam, conch and cherukuzhal.

Sources: Tholpavakoothu, Wikipedia; The Koonathara shadow puppeteers, Museum of Kerala History.

Scene 1 of 4 The invocation

ShadowStage 2025 to 2026

Computational Tholpavakkoothu, the shadow puppetry of Kerala's temples. Each puppet is a graph of joints, and a motion grammar strings poses into gestures: enter, address, strike, withdraw. A semantic light controller answers the scene, so the oil lamps behind the cloth burn steady for a quiet scene and flicker for a battle. The figures on this screen are real leather puppets, Ganesh, Rama, Ravana, Sita and Vishnu, cut out of their photographs so the lamplight comes through the punched holes as it does in the koothumadam.

Puppets: Tholpavakoothu figures in the Museu do Oriente, Lisbon, via Wikimedia Commons (CC0).

  • Joint based motion grammar
  • Semantic light control
  • Silhouette extraction
  • React · TypeScript

madhuri

off

A 1950s wooden valve radio with two knobs and five keys
The shoo shoo sign up: a username, a password, your name, and where you are.
A wooden 1970s television set
on
vol
windmill.aiEditor

Research

Computational models of how audiences watch films

I study how audiences perceive, interpret and remember cinematic media, combining computational perception models, narrative structure analysis and interactive media systems to understand how films guide attention, emotion and interpretation.

I treat attention as a perceptual mechanism and trust as a higher level interpretive judgement, and study how computational media systems shape the relationship between the two. The long term goal is artificial systems that can watch a film, find the narrative structure in it, and model viewer engagement the way a human audience does.

A wooden 1970s television set
on
vol
JOY · MRes Creative Computing, Creative Computing Institute, UAL · supervised by Professor Tim J. Smith. Upper knob is power, lower is volume.

2024 to 2025 · Research system

Multimodal viewer engagement analysis

A camera watches the audience while the audience watches the film. Facial emotion recognition, blink rate detection and an engagement curve, all synchronised to the film’s own timeline, shot by shot.

Shown here as a replay on a synthetic session.

00:00:00Shot 1 / 8

Emotion

Joy
Surprise
Fear
Sadness
Neutral

Blinks

0

0.0 per min

Engagement

0

Film timeline

In motion

Narrative DNA, running

Narrative DNA 2023 to 2026

Reading a screenplay the way a biologist reads a sequence. Every scene is tagged with its sub genre, so a film becomes a strand: where the action sits, where romance or dread takes over, how the transitions fall across the acts. A bidirectional LSTM reads the script forward and backward and predicts whether the film will succeed, at 82.7% accuracy from the script alone.

  • Scene level genre tagging
  • Sequence analysis
  • BiLSTM
  • Python

Creative work

Films, music videos, documentary

Directing and writing came before the research, and they are where the interest in narrative structure started. Around fifteen short films, advertisements and music videos, in direction, assistant direction, editing and technical production.

  1. Director

    Silent Spring

    A Malayalam music video. A love story told from inside an introvert's head, shot under Section 144 during the COVID lockdown, with the production worked out remotely.

    • Quarter finalist, WideScreen Film and Music Video Festival
    • Selected for Amazon Prime UK publishing
    • Lift-Off Filmmaker Sessions, September to October 2022
    • Student World Impact Film Festival, New York

    Worth Watch Music Video, UnniVlogs Covered on regional news

  2. Director

    How Colors are Made

    A public service film and mobile short, after a line from Rachel Houston, arguing for gender as a social construct and for the diversity that follows from it.

    • Lift-Off Filmmaker Sessions, August to November 2022
    • First Time Filmmaker Sessions
    • Student World Impact Film Festival, New York
    • Best Ad Film, Kannur University Media Fest Ad Astra
  3. Director’s cut

    SKYLAB, Sleepless Nights

    A documentary on NASA's first space station, from its launch in 1973 through its missions to the night in 1979 when it came apart in the atmosphere.

  4. Director

    Akam

    Short film.

    Watch Akam on Instagram

  5. Assistant director

    Regal Jewellers

    Television advertisement.

    Watch the advertisement on Instagram

  6. Assistant director

    Revival

    A musical documentary on the Kerala floods of 2018, and on what held afterwards.

Screenplay and camp

Publications

Published and under review

Peer reviewed

  1. Monetize the Dual: A Data Analytic Approach for Native Language and Prequel Movie Popularity Analysis

    Rahulraj P V, Sanil J, Anoop V S, Asharaf S. ICDAI, Springer, 2023.

    Springer

    Predicting film popularity from linguistic origin and franchise lineage across multiple data streams.

Earlier, in physics and education

  1. Changes in Atmospheric Air Quality in the Wake of a Lockdown Related to COVID 19 in Kerala

    Rahulraj P V, Antony E. IJAES, 2022.

    ResearchGate
  2. Influence of AI in Education System

    Rahulraj P V, Antony E. TNOU, 2022.

    ResearchGate

Experience

Where the work has been

  1. Dec 2024–Sept 2025

    AI Research & Development Engineer, Digital University Kerala

    Engineered multimodal AI systems spanning large language models, computer vision pipelines and AI-assisted verification; built LLM pipelines, multi-agent workflows and retrieval-augmented generation architectures; automated data extraction and multimodal information processing for digital media and verification tasks.

  2. Jun–Sept 2025

    AI Academic Tutor, ASAP Kerala

    Structured teaching in AI, Python, machine learning and computational thinking; workshops and mentoring for students and working professionals.

  3. Jul 2024–Sept 2025

    Co-Founder, KnowLumi

    AI-driven education platform built around adaptive interview simulation; designed and deployed a multi-agent AI interviewer; led AI workflow architecture, experimentation and deployment.

  4. Jul–Dec 2023

    Research Intern, Indian Institute of Management Kozhikode

    Large language models and enterprise information systems; prompt engineering workflows and AI-assisted information retrieval.

  5. Apr–Jun 2021

    Research Intern, CSIR-National Physical Laboratory, New Delhi

    Atmospheric analysis using NASA AERONET datasets; statistical analysis pipelines and scientific reporting.

Teaching

  1. 2018–present

    Independent Mathematics & Physics Tutor

    Mentored more than 2,000 students across physics and mathematics for higher-secondary and university level, including IIT-JEE and NEET preparation.

  2. 2022–present

    Coding & AI Instructor

    Beginner-friendly sessions in Python, AI and computational thinking for students from diverse academic backgrounds.

Creative industry

Collaborated on approximately 15 short films, advertisements and music video productions across assistant direction, creative coordination, editing support and technical production, coordinating multidisciplinary pipelines and translating between creative and technical teams.

Education

Physics, then AI, then film

  1. 2025 to now

    MRes Creative Computing, Creative Computing Institute, UAL

    On a competitive scholarship of twenty five thousand pounds. Computational media, cinematic perception, narrative intelligence, multimodal AI.

  2. 2022 to 2024

    MTech Computer Science and Engineering, AI, Digital University Kerala

    Outstanding Student Award for the batch.

  3. 2021 to 2022

    PG Diploma, Data Science and Analytics, Kannur University

  4. 2019 to 2021

    MSc Physics, Kannur University

Talks and awards

Recognition

Invited talks & workshops

  1. 2026

    The Mathematics of Cinema: Patterns, Structure, and Perception in Film

    Visiting Lecturer, Creative Computing Institute, University of the Arts London. Computational approaches to cinematic rhythm, audience attention, narrative structure and perceptual modelling.

  2. 2026

    DinoSynthesis: Generative Paleo-Acoustics via Phylogenetic VAE-GAN

    Research Presenter, UAL × CHEAD Generative AI Event, London.

  3. 2026

    AI and Industry 4.0 Tools for Innovators and Entrepreneurs

    Workshop Speaker, Digital University Kerala (online), February 2026.

  4. Invited

    AI for Humans 101

    Invited Speaker, Skill Maaya, ISTE Student Chapter MBCET. AI and machine learning fundamentals for 150+ students from colleges across Kerala.

Awards & recognition

Skills

Technical expertise

  1. AI & computational systems

    LLMs, vision-language models, computer vision, multi-agent systems, generative AI, narrative intelligence, RAG, OCR, automation pipelines, behavioural experimentation, saliency modelling, perception analytics

  2. Deep learning & generative modelling

    PyTorch, TensorFlow, LangChain, CUDA, GANs, VAEs, audio synthesis, spectral analysis, latent-space modelling

  3. Data science & analytics

    Pandas, NumPy, scikit-learn, statistical analysis, EDA, predictive modelling, ML pipelines, A/B testing, hypothesis testing, time-series analysis

  4. Creative AI & media tools

    ComfyUI, Stable Diffusion, RunwayML, Midjourney, Kling AI, Pika Labs, Leonardo AI, Adobe Firefly, TouchDesigner, Blender, Three.js, Figma, Webflow, Cargo, Premiere Pro, After Effects

  5. Programming & development

    Python, R, React, Next.js, Flask, Git, SQL, API integration, workflow automation, Linux

  6. Visualisation & research tools

    Matplotlib, Seaborn, Plotly, Tableau, Power BI, Jupyter, Google Colab, Excel

Contact

Say hello

I am glad to talk about computational perception, narrative structure, AI mediated media, or anything sitting between them. Open to research collaboration and PhD conversations.

rahulrajpv.ai@gmail.com

LinkedIn GitHub Medium Instagram X Research portfolio

London. Computational media and perception research.
Dinosaur model Baryonyx Indominus Rex hybrid thing by Absolute spino, CC BY 4.0.

Elsewhere

Three rooms

The work splits three ways, so the sites do too.