hello, I'm —

Philemon Daniel

AI engineer Voice-AI builder Startup founder NLP researcher Educator Children's author

I build voice-AI that listens to children and adults — and I run a small startup turning that research into real products. By day a professor at NIT Hamirpur; the rest of the time, a builder of careful, useful things.

SCROLL
by the numbers

Two decades, one craft.

Teaching, researching, building, shipping. A few numbers I am quietly proud of.

0
Years teaching AI, NLP and embedded systems at NIT Hamirpur
0
End-users on production AI systems I designed and shipped
0
Flagship voice-AI products in active clinical & classroom use
0
Theses supervised — 3 PhD · 37 MTech · 69 BTech major projects
currently building

The project & the startup.

A research project that grew up. HappyVoiceLearn started as a question — can we listen carefully enough to a child's voice to help them grow? VoiceHeart is the answer scaling out, as a venture.

// the AI project

HappyVoiceLearn.

A 66-point developmental framework for children's speech. 6,000+ acoustic features extracted via OpenSMILE, scored against early-childhood norms — clinically validated by CMC Vellore's Developmental Paediatrics Unit.

Latency118 ms streaming
PipelineWhisper · OpenSMILE · LiveKit
DomainsSpeech · Language · Cognitive · Motor
ValidationCMC Vellore Paediatrics
v0.4 · in pilot with three schools
// the startup
Founder & CTO

VoiceHeart.

The venture taking the same listening engine to adult psychological wellbeing — 58 markers across six domains, grounded in Logotherapy, VIA Character Strengths, ACT and Positive Psychology. Quiet, careful, clinically informed.

Wellbeinglife satisfaction, affect
Meaninglogotherapy framing
CharacterVIA strengths
ResilienceACT-aligned
Relationalconnection markers
Valueselicited & tracked
early-stage · Hamirpur, India
what I work on

Seven things I love making.

From tiny embedded chips to large language models — all of it is, secretly, the same craft: making machines listen carefully and respond kindly.

01 · Natural Language

NLP for languages on the edge of disappearing.

Unsupervised neural translation, cross-lingual embeddings, treebanks and ASR for Kangri, Kinnauri Pahari and other low-resource Indic languages. A small archive of voices, kept alive in code.

02 · Voice AI

Voice systems that listen sub-200 ms.

Real-time streaming with Whisper, ElevenLabs & LiveKit. 6,000+ acoustic features via OpenSMILE for child-development screening.

03 · LLMs & Agents

Production LLMs, RAG, MCP.

10+ shipped systems · Claude, GPT, Gemini, Llama · agentic workflows over n8n.

04 · Edge AI

On-device inference for the visually impaired.

NVIDIA Jetson · object detection · depth · natural-language scene description. NVIDIA TITAN Xp grant, 2018.

05 · VLSI & Test

The chip-test rabbit hole.

SBST, BIST, ATPG, Network-on-Chip routers. Started at Sasken & Texas Instruments OMAP.

06 · Safety

Cyber-aware AI for child safety online.

Cross-platform parental control · cyberbullying / predator detection · screen-time governance across Windows, macOS, iOS, Android, iPadOS.

07 · Computer Vision

Lanes in the Himalayas.

Lane & road detection for hilly, unstructured, shadow-heavy roads. Driver-assist where the GPS gives up.

endangered languages

Words I'm trying not to lose.

A research programme on low-resource Indic languages — corpora, treebanks, embeddings, ASR baselines, and end-to-end translation pipelines.

Languages on my desk

Each tag is months of corpus collection, dependency annotation, and patient translation. The marked ones are formally classified as definitively endangered.

Kangri Kinnauri Pahari Hindi Punjabi Bengali Tamil Marathi Gujarati English
UD
Universal Dependency treebank for Kangri — first of its kind
v0.1
Kinnauri-Pahari parallel + monolingual dataset and embeddings
Q1
Top-tier Engineering Applications of AI · Elsevier (2025)
off the clock

What I do for fun.

A scattered shelf of joys. Hover any card to bring it forward.

01 · YOUTUBE

A weekly art-integrated learning channel.

Bilingual English/Hindi episodes for school-age learners. Foundational science, drawn out loud.

🎨
02 · BOOKS

Twelve children's books, two languages.

A series exploring foundational science, written for low-resource bilingual classrooms.

📚
03 · COMMUNITY

Two open AI-education communities.

Students AI Plus & Educators AI Plus. People learning, in public, together.

🌱
04 · COUNSELLING

Listening to students, since 2015.

Member of the Student Counseling Committee at NIT Hamirpur. Mostly: just being there.

🫶
05 · HACKING

Hackathons for the love of it.

2nd place, Claude Code Hackathon 2025. 3rd place, n8n automation hackathon 2025.

the bilingual shelf

Twelve little books for curious kids.

A series of bilingual (English/Hindi) science readers, written and illustrated to make foundational STEM feel like story-time.

Why does the sky change colour?
रंग बदलता आसमान
VOL · 01
A tiny seed's big plan.
नन्हे बीज की कहानी
VOL · 02
Where do shadows go?
परछाइयाँ कहाँ जाती हैं
VOL · 03
How rivers find the sea.
नदी का सफ़र
VOL · 04
Who lives in a drop of water?
पानी की बूँद
VOL · 05
The moon's borrowed light.
चाँद की रोशनी
VOL · 06
Why does iron rust?
लोहा क्यों जंग खाता है
VOL · 07
How birds find their way home.
पक्षी का घर
VOL · 08
What is electricity, really?
बिजली क्या है
VOL · 09
A friendly tour of magnets.
चुम्बक की दुनिया
VOL · 10
Stars are very far suns.
तारे और सूरज
VOL · 11
Why do we sleep?
हम क्यों सोते हैं
VOL · 12
talks & workshops

Things I've talked about recently.

Workshops, keynotes, and hands-on sessions — for educators, students, and corporate teams.

2025
Teachers' AI Toolkit
200+ EDUCATORS · WORKSHOP
2025
Maximising Claude Learning Mode
SCHOOL · STUDENT LITERACY
2024
Decoding AI for Business Leaders
CORPORATE · TRAINING SERIES
2024
Building Custom Chatbots for Enterprise
INDUSTRY · WORKSHOP
2024
Human Values in the Age of AGI
ETHICS · CONFERENCE
working with

Partners & collaborators.

NVIDIA
Anthropic
Google for Education
ACM
CMC Vellore
NIT Hamirpur
ARM Holdings
Oxfordshire Council
say hello

let's build
something kind.

Whether it's a research collaboration, a voice-AI prototype, an AI workshop for your school or company, or just a coffee — I'd love to hear from you.