AMAdarsh MishraAI Engineer & Product Owner
Let's talk
Now · AI Engineer & Product Owner @ Capgemini

I design & shipLLM appsthat people use.

I'm Adarsh, an engineer who also owns the roadmap. At Capgemini I work on both sides of AI products: I write the model code and I run the backlog. My track record includes a 90% faster fault-triage workflow at Airtel.

Explore my workRésumé
  • Gurugram, India
  • --:--:-- IST
Portrait of Adarsh Mishra
ID://adarsh.mishraAI × PO
scroll
Python PyTorch LLMs RAG GenAI Flask Django React Next.js SQL Tableau Docker AWS Selenium Agile Product Ownership
0%

Faster fault triage

3–5 hrs → 29 min at Airtel Business

0

Internal tools shipped

Used daily by Field Ops, NOC & Product

0

Teams aligned

On one shared OPeX metrics framework

0%

Less grading effort

Assignment Feedback System, 20+ beta users

About

Half engineer, half product owner. Fully accountable for outcomes.

The short version

I started in electronics, moved into software at IIT Bombay, and learned at Airtel how good tooling changes the way a whole team works. Now I build AI systems at Capgemini, and I own the backlog that decides what they do.

Many AI projects stop at a demo. Mine have to work in production too: I define the problem, measure it, ship the model behind it and keep going until the number moves.

~/adarsh — zsh
$ whoami
role:     AI Engineer + Product Owner
at:       Capgemini
prev:     APM @ Bharti Airtel
lab:      IIT Bombay · Nano-bios
stack:    python · llms · sql · react
mode:     "ship → measure → iterate"
$

Currently

AI Engineer & Product Owner

@ Capgemini

Working on GenAI solutions and owning the delivery backlog for them.

Foundation

B.Tech, ECE

GNDEC Ludhiana · CGPA 8.25 · 2021–25

What I do

Three skills, one owner.

Most teams split these across three people and lose context at every handoff. I do all three, so nothing gets lost between them.

01

AI Engineering

LLM applications, retrieval pipelines, conversational and multimodal AI, built in Python and shipped as real products, not notebooks.

  • GenAI
  • RAG
  • Agents
  • NLP
  • Multimodal
02

Product Ownership

Backlogs, roadmaps and acceptance criteria that engineers trust. I connect stakeholders, users and data to decide what gets built and why.

  • Backlog
  • PRDs
  • Roadmaps
  • Scrum
  • OKRs
03

Automation & Data

I automate the workflows people hate and measure what changes. I set up dashboards, KPIs and SLA signals so teams can make decisions faster.

  • SQL
  • Tableau
  • Selenium
  • REST APIs
  • KPIs

~/dev

Don't just read about it. Run it.

This is how I think, written as code. The terminal is a real shell: type help, neofetch or sudo hire-me.

adarsh.ts — portfolio — Visual Studio Code
portfolio › src › adarsh.ts
1234567891011121314151617
main 0 problemsLn 1, Col 1UTF-8TypeScript● typing…
adarsh@portfolio: ~ (zsh)120×32

Last login: just now on ttys001

Welcome to adarsh.dev. This is a real shell. Type 'help' to begin.

Experience

From circuits to copilots.

  1. AI Engineer & Product Owner

    Capgemini

    2025 — Present
    • Own the product backlog for AI initiatives, turning business problems into epics, user stories and clear acceptance criteria.
    • Design and build GenAI / ML solutions, including LLM-powered apps, retrieval pipelines and conversational interfaces, from prototype to production.
    • Bridge client stakeholders, data scientists and engineers through sprint planning, reviews and demos that keep delivery tied to outcomes.
    • GenAI
    • LLMs
    • Product Ownership
    • Agile

  2. Associate Product Manager, Automation & Analytics

    Bharti Airtel · Airtel Business

    Jan 2025 — 2025
    • Cut B2B network fault investigation from 3–5 hours to 29 minutes (90%+) with a fault-to-service correlation engine.
    • Owned an internal analytics platform used daily by Field Ops, NOC, Product and Automation teams.
    • Aligned 4 cross-functional teams on a shared OPeX metrics framework and pitched an AI-augmentation roadmap to senior leadership.
    • Product Spec
    • Python
    • Selenium
    • Tableau
    • SLA Analytics

  3. Software Development Intern

    IIT Bombay · Nano-bios Lab

    Mar 2024 — Jun 2024
    • Shipped a research collaboration platform (Flask + YouTube API) in 12 weeks. The lab adopted it as its primary async tool.
    • Built 2 Django healthcare modules with RBAC, cutting researcher data-retrieval time by ~40%.
    • Acted as the sole interface between 5 scientists and engineering on a 2-week release cadence.
    • Flask
    • Django
    • RBAC
    • User Research

  4. Research Intern

    Semi-Conductor Laboratory (ISRO), Mohali

    2023
    • Ran experimental analysis of MOS capacitors for oxide furnaces and studied VLSI fabrication workflows.
    • VLSI
    • Experimental Analysis

  5. B.Tech, Electronics & Communication · CGPA 8.25

    Guru Nanak Dev Engineering College

    2021 — 2025
    • Overall Coordinator for APEX Tech Fest (200-member committee). Grew club active membership 3× as Student Coordinator.
    • Pitched a healthcare AI concept at BUILD'23 Global Hackathon and drew funding interest from an MNC sponsor.
    • Networks
    • DBMS
    • Leadership

Selected work

Shipped, measured, still in use.

Airtel Business

Network Fault Impact Engine

3–5 hrs → 29 min

Mapped a high-friction telecom ops workflow, wrote the spec, designed REST API contracts and shipped automation that correlates network faults with impacted B2B services.

  • Python
  • Selenium
  • REST APIs
  • Tableau
Read case study

Deep learning research

I2S-TFCKD Speech Enhancement

Teacher → student distillation

Reference implementation of intra/inter-set knowledge distillation with time-frequency calibration. DPDCRN teacher and student, MR-STFT + L1 losses, AMP and TensorBoard.

  • PyTorch
  • Knowledge Distillation
  • Audio
  • AMP
Source

Conversational AI · IIT Bombay

Video Analyzer & TranscriptBot

Chat with any YouTube video

Conversational AI over YouTube transcripts. Ask questions, get grounded answers. It grew into the lab’s research collaboration platform and replaced scattered email threads.

  • Django
  • Flask
  • NLP
  • YouTube API
Source

Multimodal ML

CLIP Multimodal Exploration

Zero-shot image ↔ text matching

Hands-on exploration of contrastive language-image pretraining: zero-shot classification and matching text snippets to images.

  • Python
  • PyTorch
  • Computer Vision
  • Embeddings
Source
12projects in totalSee all work

Featured case study · Airtel Business

Network fault investigation: from 3–5 hours to 29 minutes.

Analysts were matching network faults to impacted enterprise customers by hand. I treated it as a discovery problem: I mapped the workflow, studied 3 months of incidents, aligned 4 teams on what “impact” meant, wrote the spec and shipped the automation.

Read the full breakdown
before
3–5 hrs
after
29 min
Fault detectedAuto impact mapShared KPI viewProactive SLA signal

Operating principles

How I build AI products.

  1. 01

    Workflow > feature list

    Good products start with how people actually work, not with a list of features.

  2. 02

    Trust before automation

    An AI system earns adoption by being right and explainable before it gets clever.

  3. 03

    Metrics settle debates

    When teams disagree on success, a shared number settles it faster than a meeting.

  4. 04

    Decision time is the KPI

    Internal tools win when they shrink time-to-decision, not just manual effort.

Stack

Tools I use to get from idea to impact.

AI / ML

  • Generative AI
  • LLM Applications
  • RAG
  • Prompt Engineering
  • NLP
  • Computer Vision
  • Multimodal (CLIP)
  • Knowledge Distillation
  • PyTorch
  • scikit-learn
  • Model Evaluation

Product

  • Product Ownership
  • Backlog Management
  • PRD Writing
  • Roadmapping
  • User Research
  • Stakeholder Management
  • OKRs & KPIs
  • Agile / Scrum
  • A/B Testing

Engineering

  • Python
  • Flask
  • Django
  • REST APIs
  • React
  • Next.js
  • Node.js
  • Selenium
  • Docker
  • Git
  • Linux
  • AWS

Data

  • SQL
  • MongoDB
  • Pandas
  • NumPy
  • Tableau
  • SLA / MTTR Analytics
  • Dashboards