Skip to content
Hitansh Gopani

Systems onlineMumbai → Global

Hitansh Gopani — AI/ML EngineerI build AI systems that keep working after the demo is over.

Turning emerging technology into systems, products and workflows that solve real problems — as an AI/ML Engineer at Pitch Perfekt Collective and co-founder of LightNoteAI.

  • AI systems
  • LLM evals
  • Agents
  • GenAI
  • Automation
Explore work →Get in touch

eval · hitansh

  1. Ships working softwareLive demos: Crucible, VoxGate, Sigmanot run
  2. Evaluates before trustingTrustLLM prototype · honest baselines in Incident Intelnot run
  3. Documents what broke14 findings logged in Cruciblenot run
  4. Leads & builds teamsCo-founder · 15+ devs mentorednot run
  5. Explains clearlyTrained professionals & students · Medium articlesnot run
Automated tests in Crucible
Automated tests in Crucible Verify (opens in a new tab)
Developers mentored
Developers mentored

Explore as

The 30-second version

  • AI/ML Engineer at Pitch Perfekt Collective (Jul 2026–) · Co-founder, LightNoteAI (Oct 2025–)
  • Prior: AI/ML Trainer at QUASTECH (professionals & final-year students), Radical AI (AI software, remote), Vendiman (data)
  • Strengths: LLM systems & evaluation, agents, voice, vision ML, shipping to production
  • Open to conversations · Mumbai · remote-friendly, globally

01 · Why I build

Building AI isn't the hard part.Keeping it working is.From data,to intelligence,to production.

Hitansh Gopani in a black suit, arms crossed

Hitansh Gopani

AI/ML Engineer · Mumbai, India

About the model

Engineer. Builder. Thinker.

I'm an AI engineer in Mumbai. By day I build production AI pipelines at Pitch Perfekt Collective; I also co-founded LightNoteAI, the intelligence layer for proposals.

I care about the unglamorous part — evaluation, failure modes, tests and logs — because that's what decides whether an AI system is still trusted a month after launch.

I learn in public: 15+ developers mentored, articles on Medium, and a habit of writing down what broke.

Engineer

  • AI / ML
  • LLM systems
  • Evaluation

Builder

  • AI products
  • Agents
  • Automation

Thinker

  • Systems
  • Architecture
  • Strategy

↓  Global impact

Value

What I bring to the table

  • 01

    Technical depth

    Across the AI stack — data pipelines, classic ML, LLM apps, RAG, agents, APIs and deployment.

    4 end-to-end builds with public code · 1,200+ tests in Crucible

  • 02

    Systems thinking

    I don't treat AI as an isolated model. I design how it connects to data, infrastructure, people and workflows.

    System map above

  • 03

    Builder energy

    I prototype, integrate and ship — and when a demo crashes 12 hours before a deadline, I rebuild it.

    Avishkar zonal win · Code Odyssey runner-up

  • 04

    Business orientation

    I don't only build inside existing systems — I find problems and turn them into products. The goal isn't clever technology; it's something useful.

    LightNoteAI · McKinsey.org Forward

  • 05

    Global mindset

    Based in Mumbai, building for teams anywhere — from compliance packs for the UAE to remote AI work.

    VoxGate KYC-UAE pack

Intended use

The system I build

Every project I take on lives somewhere on this map. Hover a module to see the tools, and where it's already proven.

module · evals

Evals & guardrails

Hallucination, grounding, consistency and toxicity checks before anything reaches users.

  • Grounding
  • Hallucination
  • Toxicity
  • Audit logs

Proven in

Out of scope · Demos that only need to work once.

Benchmarks

Selected work

Each project: the problem, the demo version, and what it took to reach production — plus what broke, where I've written it up. Flip a card for details.
  • In progressLLM Systems & Evals

    TrustLLM

    Test your LLMs before your users do.

    Prototype · up to 4 models · 5 metrics + composite (simulated outputs)

    DemonstratesLLM evaluation · Guardrail design · Product thinking

    Case study →
  • LiveVerifiable AI

    Crucible

    Every fine-tuned model gets a birth certificate.

    1,200+ tests · 14 documented findings

    DemonstratesSystems engineering · Testing discipline · Debugging under pressure

    Case study →
  • LiveVoice & Agents

    VoxGate

    Voice interviews that gate compliance decisions — explainably.

    Pack architecture · KYC-UAE launch pack

    DemonstratesAgent design · Voice interfaces · Regulated domains

    Case study →
  • ShippedVision & ML

    Incident Intel

    Spot what's unusual in fixed-camera footage — and say how sure you are.

    3 approaches compared · baseline AUC 0.709, deployed autoencoder 0.674

    DemonstratesML rigour · Data versioning · Honest metrics

    Case study →

Leadership · Co-founder · Oct 2025 – present

I co-founded LightNoteAI to help businesses win more proposals.

Co-founded in Oct 2025, during the final year of my B.Tech. The intelligence layer for proposals.

  1. The problem

    Proposals reach clients without anyone checking their quality, their weak spots, or what the buyer will actually evaluate.

  2. What we're building

    An AI proposal-intelligence platform that analyses proposal quality, finds weaknesses, surfaces buyer and evaluation signals, and recommends fixes — before the proposal goes out.

  3. What I work across

    • Product
    • AI
    • Strategy
    • Growth

    Focus areas

    • Proposal scoring
    • Win intelligence
    • LLM workflows
    • SaaS infrastructure
  4. Visit lightnoteai.com (opens in a new tab)

What co-founding proves

  • 01

    Initiative

    I didn't wait for a job title to build a product. I spotted a problem and co-founded a company around it while still a student.

  • 02

    Leadership & ownership

    I work across product, AI, strategy and growth — the decisions, not just the tickets.

  • 03

    Real problem-solving

    Proposal quality is a business problem with money on the line. The AI has to change an outcome, not just produce text.

  • 04

    Commercial thinking

    Product strategy and go-to-market sit next to the model work. Building something businesses will adopt is its own discipline.

The lab

I like building things that shouldn't exist yet.

Some started as experiments. Some became products. Some broke spectacularly. All of them taught me something. Open a row for the details.
Earlier builds · 2023–2025 ↗StackFlow · Solana staking analyticsDaoConnect · Polkadot DAO governance · hackathonSmartScaffold · Metis dApp toolkit

Training data

Experience

Everything I've worked on, newest first — with co-founding LightNoteAI highlighted as leadership. B.Tech in Information Technology, K. J. Somaiya Institute of Technology (2022–2026).

  1. Jul 2026 – present

    Artificial Intelligence Engineer · Pitch Perfekt Collective

    • Production AI pipelines that keep creative assets, characters and workflows organised, reusable and ready for production.

  2. Dec 2025 – Feb 2026

    Training · Communication

    AI/ML Trainer · QUASTECH

    • Trained working professionals and final-year students in AI/ML — for real projects and placement readiness.
    • Designed and delivered the curriculum on ML, transformers and GenAI; mentored learners through their projects.
    • Kept content current as AI tools and industry use cases changed month to month.

  3. Oct 2025 – present

    Leadership

    Co-Founder · LightNoteAI

    The intelligence layer for proposals. I work across product, AI, strategy and growth.

    What co-founding it proves →

  4. 2024 – 2025

    Mentorship

    Facilitator · Google Cloud Arcade

    • Co-ran a mentorship community for Google Cloud Arcade learners.

  5. Jun – Jul 2024

    Contributor · GirlScript Summer of Code

    • Open-source contributions during GSSoC'24.

  6. May – Sep 2024

    Data Analyst · Vendiman

    • Automated SQL reporting and dashboards in Metabase.
    • Built WhatsApp automation workflows.

  7. Nov 2023 – Apr 2024

    AI Software Developer · Radical AI

    • Built AI-driven education apps in Python and web tech; open-source work on access to AI.

  8. Sep 2023 – Jul 2024

    Team leadership

    Technical Manager · IoT Cell, KJSIT

    • Led the build of a full e-commerce platform with Stripe payments.
    • Mentored 15+ developers on building and deploying full-stack apps.

  9. Jun – Sep 2022

    Outreach Coordinator · Google Developer Student Clubs, KJSIT

    • Outreach for community events including “Sprint”.

In print

  • Tejas Magazine 4.2 — “Smart Bharat”

    Published article

In the room

Ethical considerations

What I care about

  • Value 01

    Dependable over impressive

  • Value 02

    Own it when it breaks

  • Value 03

    Teach to understand

  • Value 04

    Build for access

  • Value 05

    Stay in rooms where you're not the smartest

Limitations

Where I'm still growing

Every good model card says where the model falls short. So does this one.
  • Three months into a full-time production role — my longest runs are in projects and a startup, not a large team.
  • Strongest in LLM systems, evaluation and agents; still growing in large-scale infrastructure and distributed training.
  • Several projects are early-stage and not yet deployed publicly — those show code and a case study, not a live demo.

Changelog

Build log

  1. v3.0In training.
  2. v2.3Sep 2026Dell Technologies Forum, Mumbai — notes on moving AI from impressive to dependable.
  3. v2.2Jul 2026AI/ML Engineer at Pitch Perfekt Collective. CSCMP Generative AI for Supply Chain certificate.
  4. v2.1May 2026Graduated — B.Tech IT, K. J. Somaiya Institute of Technology.
  5. v2.0Mar 2026Snowflake “Zero to Agents” — guided hands-on workshop: a governed agent with text-to-SQL and RAG.
  6. v1.9Dec 2025AI/ML Trainer at QUASTECH · Zonal winner at Avishkar · McKinsey.org Forward.
  7. v1.8Oct 2025Co-founded LightNoteAI.
  8. v1.7Sep 20251st Runner-Up, Code Odyssey 4.0 (Tata STRIVE × KJSIT) — SapnaForge.
  9. v1.52024 – 2025Google Cloud Arcade facilitator · workshops and mentoring.
  10. v1.02023 – 2024First AI internship (Radical AI, remote); led IoT Cell; Vendiman data internship; Postman Student Expert.
  11. v0.12022Started at KJSIT; GDSC outreach.

Writing

Awards & certifications

  • 1st Runner-Up — Code Odyssey 4.0 (Tata STRIVE × KJSIT)
  • Zonal Winner — Avishkar
  • Snowflake — Zero to Agents
  • CSCMP — Generative AI for Supply Chain (2026)
  • McKinsey.org Forward Program (2025)
  • Lean Six Sigma Foundations (PMI, 2025)
  • Postman Student Expert (2023)

Research interests

Questions I'm exploring

Trajectory

Where I'm going

  1. Now

    AI/ML Engineer

    Building production AI pipelines at Pitch Perfekt Collective; co-founding LightNoteAI.

  2. Next

    AI Systems Engineer

    Owning whole systems end to end — data, models, evaluation, infrastructure and the people who use them.

  3. Long term

    AI & emerging-tech strategist

    Deciding what should be built, and why — with the engineering depth to know what it will take.

Ask Hitansh · the intelligence layer

Ask about my work

Every node below is real — a skill, strength, project, role or research question from this site, linked the way the work actually connects. Ask a question and watch retrieval light up the evidence; click a node to ask about it.

Retrieval over this site's verified content · runs in your browser · no LLM, nothing leaves the page.

Console

Prefer a terminal?

The whole site, as a shell. Try whoami, ls projects or open crucible.
hitansh.os — zsh
hitansh.os — type help, or tap a command below.
hitansh@lab:~$ whoami
Hitansh Gopani — AI/ML EngineerI build AI systems that keep working after the demo is over.Mumbai, India · open to conversations

How to reach this model

Let's build something that survives production.

Email me →GitHub(opens in a new tab)LinkedIn(opens in a new tab)Medium(opens in a new tab)X(opens in a new tab)

Based in Mumbai · building for a global future