Natesh Aravind S
Work
About
Resume

Hey explorer, you’ve reached

Natesh Aravind S

Natesh is an IIT-educated researcher and innovator, an NID-trained systems thinker and service designer, a founding product hire building agentic AI for healthcare and an astrophysics lover, storyteller and brand builder.

LinkedIn ↗

In one line

Engineer by training, designer by method, builder by instinct, connecting unconventional dots across engineering, design and business.

Now

Founding product hire at Pranik.ai, Mondee’s AI health-tech venture

Sep 2025 to now (1+ year)

Orbiting topics: Product, Field research, Service design, Systems thinking, Conversation design, AI evals, Behavioral science, Agentic workflows.

  • 20,000+ consultations on the platform
  • 100+ doctors on the product
  • 7 Indian languages
  • ~80% less documentation time
  • ~70% fewer drug-name errors
  • 2+ min → ~35 s prescription latency
  • 400+ patients in field research
  • 20+ hospitals, clinics and camps
  • 30+ agent workflows in a living PRD
  • DAT AIR 1 NID entrance exam
  • Class A CDSCO test license

Studied, worked and collaborated with

  • IIT Madras
  • IIT Tirupati
  • ISRO
  • National Institute of Design
  • Mondee
  • Tata Elxsi

(01)In brief

Hey explorer! Here’s the 60-second version.

I’m Natesh, an engineer turned design-driven product builder. I like solving complex socio-technical problems where design, engineering and business meet. That range lets me connect unconventional dots, and often spot the insight that would otherwise surface only after real-world testing, or after hitting a roadblock.

01Who
Founding product hire at Pranik.ai, Mondee’s AI health-tech venture, reporting to the CEO and CTO. 3+ years of experience across AI products, engineering R&D and applied design strategy.
02What I’ve shipped
Two agentic AI healthcare apps, one for doctors and one for patients, from 0 to 1 (now in UAT). The doctor app has crossed 20,000+ consultations with 100+ doctors.
03What I go deep on
AI evals and clinical safety, conversation design for voice agents, field research, and building my own tools with Claude Code and Claude Design.
04How I’m wired
Mechanical engineer (IIT Tirupati), then EV R&D (Tata Elxsi), then strategic design (NID Ahmedabad, DAT AIR 1), now AI product. I think in systems and test in the field.
Resume
The one for your ATS. Download PDF ↓ View it here →

(02)Selected work

Selected work.

Pranik.ai, Mondee’s AI health-tech venture. Two agentic apps, one for doctors and one for patients, voice-first in 7 Indian languages. Sorted by the hat I wore.

Hat 01

AI evals, models and clinical safety

Rank AI errors by clinical risk, not by how often they happen.

1 case study

  1. 01 Ranking AI errors by clinical risk: the evals loop behind a medical AI. Every doctor correction, from diff to root cause to the right team. ~70% fewer drug-name errors, to ~10% of mentions 2025–now

Hat 02

Product leadership

Set up the product function, then took two agentic apps from 0 to 1.

3 case studies

  1. 02 Taking agentic AI healthcare from 0 to 1, for doctors and for patients. First product person. Two squads, two apps, 20+ hospitals and camps. ~90% of active pilot doctors stayed for phase two 2025–now
  2. 03 Agent Flow Blueprint: a spec format for products that branch on conversation. A living spec for agents that branch on conversation, built in Claude Code. 30+ agent workflows in one source of truth 2026
  3. 04 Teaching an AI twin to practice like its doctor. How any doctor can train their AI twin without our team in the room. 25 doctors with a live twin in UAT 2026

Hat 03

Design and UX research

Sit with the patients first. Then change how the AI asks.

2 case studies

  1. 05 “Side unclear”: turning a field observation into conversation design. Patients pointed at where it hurt; the AI guessed a side. So it learned to ask. Under 10% wrong-side errors in camp testing 2025–26
  2. 06 Rolling out AI into 20+ hospitals, clinics and health camps. Before the product could help a doctor, it had to fit an OPD. 10+ → <2 min documentation time at Bhaktivedanta 2025–26

Some details are generalized or left out for confidentiality. All work →

(03)How I work

Eight habits I keep coming back to.

Six from Pranik.ai, two from NID projects

  1. 01Pranik.ai

    Rank errors by harm, not by count.

    A left-right flip is a patient-safety failure. A misspelled symptom is harmless, however often it shows up. The evals loop sorts by clinical risk first.

  2. 02Pranik.ai

    Kill things on evidence.

    We built a camera layer so the AI could see where patients pointed. It flipped sides and slowed everything down. We went back to audio and better questions.

  3. 03Pranik.ai

    Sit with users before writing specs.

    An elderly woman cried while telling our AI avatar about her problems. It showed how much people trust it, and pushed us to build empathy into how it speaks.

  4. 04Pranik.ai

    If the spec format is broken, rebuild the format.

    PRDs and user stories don’t describe agents that branch on conversation. So I built a living blueprint the whole team works from.

  5. 05Pranik.ai

    Patch the UX this week, fix the model next month.

    Most AI issues have a fast product fix and a slow data fix. I ship the first while data science builds the second.

  6. 06Pranik.ai

    Ask “noticed any change?” before saying what changed.

    I validate fixes with doctors blind, then check their answer against the evals data.

  7. 07NID project

    Design for people at their most anxious.

    At a GSRTC bus depot, we rode with commuters through rush hour. Anxious riders skipped a 7-step boarding sequence to grab a seat, and a simple boarding-pass cue restored it in a campus simulation. It’s the same problem as a worried patient talking to a health AI.

  8. 08NID project

    Zoom out to the whole system, then zoom in on the process.

    For Oizom’s CEO, we ran 31 stakeholder interviews, from CXOs to team members, and mapped the company as one system. Client issues were bouncing between six teams, so I designed a delegation workflow: a support team handles issues, and only new or critical ones reach R&D.

(04)Trajectory

A nonlinear path, plotted on purpose.

T+ 2018–22 · Stage 01/05

01Launch · 2018–22

IIT Tirupati

B.Tech, Mechanical Engineering

Engineering foundations, first research, first teams to lead.

What I took from it

  • First principles and research rigor: how machines and materials behave, and an ISRO-funded thesis on twin-wire arc additive manufacturing.
  • Building with a team against a deadline: an autonomous farming robot that took bronze at the Inter IIT Tech Meet.
  • Leading people and budgets early: as Technical Affairs Secretary I ran a 150+ member council and a ₹40L budget, and I was on the founding team of the IDEAS² innovation lab.
  • Storytelling and giving back: co-founding Cosmos And I, an astronomy magazine, and guiding 1,500+ government-school students on careers.

Payload bay · 8 skills loaded

  • Mechanical design
  • Robotics
  • Research methods
  • MATLAB
  • Student leadership
  • Team building
  • Storytelling
  • Mentoring

(05)Skills

What I bring to the table.

Everything the trajectory loaded into the payload bay, sorted into seven skill sets and the toolkit I work with.

01

Contextual first principles

Questioning the right thing at the right time, informed by context, research and curiosity.

  • Root-cause analysis
  • Engineering rigor
  • Systems thinking
  • Problem framing

02

Humans, brain and behavior

Behavioral science turned into product decisions, tested in the field.

  • Fogg
  • COM-B
  • AEIOU
  • Ethnography
  • UX research
  • Behavior simulations

03

AI-powered builder

Shipping AI products that have to be right, and building my own tools with AI.

  • LLM evals
  • Dataset preparation
  • Prompt and context engineering
  • RAG
  • Voice AI
  • Guardrails
  • Human in the loop
  • Model trade-offs: quality, latency, cost
  • Agentic workflows
  • AI prototyping

04

Product leadership

Owning the problem from discovery to rollout, with the team.

  • 0 to 1
  • PRDs and specs
  • Roadmaps
  • Prioritization
  • Product experiments
  • Retention metrics
  • Squads and SOPs

05

Service and systems

Seeing the whole service, and the system behind it.

  • Personas
  • Journey maps
  • Service blueprints
  • Systems mapping
  • Success metrics
  • Latent opportunities

06

Branding and storytelling

Finding the story, then giving it a shape people remember.

  • Brand identity
  • Positioning and messaging
  • Design systems
  • Pitch decks
  • Video scripts and storyboards
  • GTM

07

Leadership and influence

Working across organizations, regulators and communities.

  • Stakeholder engagement
  • Change management
  • Business cases
  • CDSCO, ABDM, DPDP

08

Toolkit

What I build, design, track and ship with.

  • Claude Code
  • Claude Design
  • Figma
  • Framer
  • Notion
  • Jira
  • Google Analytics
  • PostHog
  • SQL

(06)Mission archives

Mission archives.

Campus leadership, NID projects, hardware side projects, and what I do off the clock.

Campus leadershipIIT Tirupati

  • Technical Affairs Council2021–22 Secretary of a 150+ member council: secured and managed a ₹40L budget across 8+ tech clubs, and started the first intra-college tech fest.
  • IDEAS² Innovation Lab2019–21 Head and founding committee member: set up an innovation lab on the IIT Madras CFI model, with institute funding won from the director and registrar.
  • Learning by Doing V22020–21 A hands-on robotics learning platform with Prof. T S Natarajan: designed it, coordinated 20+ interns, and pitched it to IIT Madras leadership.

NID projectsNID Ahmedabad

  • It’s late and I’m anxious2024 Cognitive tunnelling at a GSRTC bus depot, and an information-design fix. A sole-authored white paper.
  • Oizom2024 An organization audit for an IoT startup’s CEO, from 31 stakeholder interviews.
  • Sree Sai Hyderabadi Biryani2024 Positioning a third outlet through consumer psychology and a brand story.
  • Beyond the Clock2025 Understanding overtime in India’s IT industry. Published at ICoRD 2025 (Springer).
  • Water policy study Non-perennial river sources and how communities relate to water, with Prof. Hirter (HSLU, Switzerland).
  • Design in India, Design for the World2025 A symposium at NID on how strategic design can feed the 2047 Viksit Bharat vision.

Hardware side projectsRobotics and design

  • AMOS AutoMicroUAS2021 Final-year internship: a hexacopter mainframe and a foldable arm connector that doubled the arm’s rotation to 180°.
  • Transparent N95 mask2021 A lab project with Dr. D V Kiran: redesigned for ergonomics and breathability with 3D-printed molds. Amara Raja took it to a small pilot production run.

Off the clockSpace and giving back

  • Cosmos And I2019–20 Co-founded an astronomy magazine: 4 editions, 1,000+ reads.
  • Career guidance2018–22 Volunteered with 1,500+ government-school students.
  • Interests Setting up telescopes for stargazing, trekking, camping, cycling, and Dan Brown-style thrillers.
Pending launch