AI Project Manager

The AI project manager who has actually shipped AI.

Not just managing AI projects from a spreadsheet — since 2016 I have built custom NLP models, architected LLM pipelines, shipped GenAI platforms to production, and led conversational AI delivery serving 2M+ users. That is the difference between an AI project manager and one who truly understands the stack.

9+

Years in AI delivery

2M+

Users on a single AI product

1M+

Peak daily AI conversations

100%

On-time AI delivery

The role

What an AI PM does differently.

Managing an AI project is fundamentally different from managing a traditional software project. AI systems are probabilistic, not deterministic. Scope evolves as models are trained and tested. Evaluation frameworks replace conventional QA checklists.

A good AI PM understands the technology deeply enough to manage it accurately — to know when a model is underperforming, when a RAG architecture needs redesign, and how to set realistic expectations with business stakeholders who've read the hype.

I have been building AI products since 2016 — before the term "GenAI" existed. That depth is what I bring to every AI programme I lead.

Traditional Project

AI / GenAI Project

Deterministic output

Probabilistic, non-deterministic output

QA via test cases

Evaluation frameworks + human review

Fixed technical scope

Scope evolves with model performance

Standard defect types

Hallucinations, drift, bias, latency

Dev team builds it

Prompt engineering + ML engineering

UAT signs it off

Red-teaming + production monitoring

Core capabilities

What I deliver on AI programmes.

01

AI Delivery Lifecycle Management

Managing the unique lifecycle of AI products: from data requirements and model selection through prompt engineering, evaluation, human-in-the-loop design, and production deployment.

  • AI project scoping & feasibility
  • Model selection & vendor evaluation
  • Prompt engineering oversight
  • Evaluation framework design
  • Human-in-the-loop architecture
02

LLM & GenAI Orchestration

Architecting and overseeing multi-model LLM pipelines with dynamic routing, RAG integration, cost optimisation, and latency SLA management for enterprise production deployments.

  • OpenRouter multi-model routing
  • RAG pipeline design & oversight
  • LangChain integration delivery
  • LLM cost & accuracy optimisation
  • Context window management
03

Conversational AI Delivery

End-to-end delivery of chatbot and voicebot solutions — from intent taxonomy design and NLU training through channel integration, human handoff, and production monitoring.

  • Intent & entity architecture
  • NLU training & evaluation
  • Multichannel bot deployment
  • AI-human escalation models
  • Persona & conversation design
04

AI Risk & Governance

Managing the unique risks of AI projects: hallucinations, model drift, bias, data privacy, regulatory compliance, and the balance between accuracy and latency in production.

  • AI risk register management
  • Hallucination mitigation strategy
  • Data governance & compliance
  • Responsible AI frameworks
  • Model monitoring & alerting
05

Cross-Functional AI Team Leadership

Bridging the communication gap between business stakeholders, data scientists, ML engineers, software developers, and QA teams in AI-first product delivery.

  • AI team structure & hiring
  • ML engineer collaboration
  • Stakeholder education on AI
  • AI sprint planning & ceremonies
  • Research-to-production handoffs
06

AI Quality & Evaluation

Designing rigorous evaluation frameworks for AI systems — beyond traditional QA — including human evaluation pipelines, benchmark testing, and continuous monitoring.

  • LLM evaluation framework design
  • Human-in-the-loop testing
  • Production model monitoring
  • Accuracy & performance benchmarking
  • Red-teaming & safety testing

AI track record

9 years of AI product delivery.

2025–26

Enterprise GenAI Platform — Product Manager

Zero to production MVP. Architected multi-model LLM routing via OpenRouter. Implemented dual-channel AI + human resolution model. 100% sprint delivery consistency. Full SDLC from scratch.

2022–25

GenAI & Conversational AI Portfolio — Program Manager

Led GenAI and Conversational AI workstreams across global enterprise clients. Delivered production-ready chatbot and voicebot projects on Dialogflow, Rasa, and IBM Watson — 100% on-time.

2019

Sony MAX Filmykaant — Project Manager, Conversational AI

India's highest-traffic entertainment chatbot. 2M+ unique users. 1M+ conversations/day. Built custom Hinglish NLP/NLU model from scratch when no off-the-shelf solution existed. Facebook Top 10 Chatbot. FB Start selected.

2018

QuizMaster Bot — Project Manager

1.5M+ unique users on Facebook Messenger. Top 10 Chatbot globally. Selected for FB Start programme. Competitive quiz bot with real-time leaderboards and gamification features.

2016

Conversational AI Practice — Founder, Cedex Technologies

Built one of India's first dedicated Conversational AI practices in 2016. Delivered chatbot and voicebot solutions for Orange Telecom, Gillette, Adani Power, Fevicol, and GBM using IBM Watson, Dialogflow, Rasa, Twilio Flex, Alexa, and Google Actions.

Technical depth

AI technologies I've actually delivered with.

Not just familiar — production-delivered. Every technology below has been used to ship real AI products to real users.

Generative AI & LLM

OpenRouterLangChainRAGGemini APIOpenAI APIGoogle AI StudioPrompt EngineeringLLM EvaluationMulti-model Orchestration

Conversational AI

DialogflowRasaIBM WatsonAmazon LexMicrosoft Bot FrameworkTwilio FlexAlexa SkillsGoogle ActionsCustom NLP/NLU

AI Productivity Tools

ClaudeChatGPTPerplexityCursorGitHub Copilot

Building AI since 2016 · Available now

Have an AI project that needs a real PM?

Whether it's a GenAI platform, a conversational AI product, or an LLM-powered enterprise application — I bring delivery rigour and genuine AI depth to make it happen.