Product Manager · Conversational AI · New York

I design, ship, and measure conversational AI.

I'm a product manager on an enterprise chatbot that handles 300K+ sessions a day, and I build the LLM tools that show how it performs. I've spent 10+ years shipping products, from SaaS used by 40,000+ institutions to agentic AI. I prototype ideas myself before asking engineering to build them.

Currently Product Manager @ Fortune 50 Telecom · via Capgemini · Google ADK, Conversational Agents & CX Agent Studio

Connect on LinkedIn Resume (PDF)

chatbot sessions a day on the assistant I help run: 300K+; chatbot sessions a day on the assistant I help run
containment: conversations resolved without a human agent: 78%; containment: conversations resolved without a human agent
chats a day on a field-tech bot I took from 0 to launch in 4 months: 10K+; chats a day on a field-tech bot I took from 0 to launch in 4 months
internal users of the transcript analysis tool I built: 1,000+; internal users of the transcript analysis tool I built

Selected Work

Featured projects

All work

01 · Generative answer

Virtual Assistant

Online

My internet keeps dropping in the evenings. Third night in a row.

Sorry about that. I checked your connection, and here's what I found:

Connection check · last 7 days

Gateway statusOnline
Outages in your areaNone
Reconnects14 · mostly 7–10pm

An evening-only pattern with no outage usually means Wi-Fi congestion, not a line problem. Moving your network to a less crowded channel often fixes it. Want me to try that?

Optimize my Wi-Fi Restart gateway Talk to an agent

Type a message…

02 · Handoff with context

Virtual Assistant

Connecting you…

I was charged twice for my plan this month.

I can see two charges of $64.99 on Oct 3. Billing disputes need a specialist, so I'm connecting you to one now.

Shared with the agent

Customer reports a duplicate monthly charge. Two identical charges on Oct 3 confirmed. Identity already verified. No earlier contact about this.

Billing Verified Sentiment: frustrated

Alex from Billing joined the chat

Alex · Billing

Hi! I've got the details, so no need to repeat anything. I'm reversing the duplicate charge now.

Type a message…

03 · Field-tech assistant

Tech Assist

Work order #48213

Downstream SNR is 29 dB at the tap. Customer still has intermittent drops. What next?

29 dB is below the 33 dB minimum, so the problem is likely between the tap and the home. Check in this order:

Inspect the drop connectors for corrosion
Replace any splitters with more than 2 ways
Re-test SNR at the ground block

Source: Install & Repair Guide §4.2

Replaced a 4-way splitter. SNR is 36 dB now.

That's within spec. I've added the fix to the work order notes.

Close work order Run speed test

Type a message…

Enterprise Virtual Assistant

Enterprise

Product manager on a high-volume omnichannel chatbot: moving it to Google ADK and shipping its first generative AI features.

chatbot sessions / day: 300K+; chatbot sessions / day
containment: 78%; containment

Google ADKCX Agent StudioVertex AIDialogflow

Case Study

CI Conversation Insights

Last 7 days ▾ All channels ▾ Intent: any ▾

Conversations analyzed

1.42M

last 7 days

Containment

78%

▲ 0.6 pts

Negative sentiment

14%

▼ 1.2 pts

Repeat contact · 48h

9.3%

▲ 0.4 pts

Sentiment mix by day

Positive Neutral Negative

Top escalation drivers

Asked for a person on the first turn 29%
Billing dispute (needs a person) 21%
Same intent missed 2+ times 17%
Identity check failed 12%
Outage status unclear 8%

Transcripts · 5 of 2,906 flagged

C-20931 Wi-Fi drops App Negative
C-20928 Bill explanation Web Neutral
C-20925 Plan upgrade SMS Positive
C-20919 Equipment return Web Neutral
C-20914 Appointment App Positive

C-20931 · Wi-Fi drops

App · 9 turns · 6m 12s

LLM summary

Customer reported evening Wi-Fi drops for three nights. The assistant ran a connection check and suggested a channel change. The customer had already tried that and asked for a technician. The assistant offered the same fix again before escalating.

Flagged moment · turn 6

“I already did that yesterday. Can someone just come look at it?”

repeated-suggestion tech-visit-intent add to review queue

Chat Transcript Analysis Tool

Enterprise

An internal Gemini tool that scores chatbot conversations for sentiment, writes summaries and generates agent metrics. I built it hands-on with Claude and other AI coding tools.

internal users: 1,000+; internal users
daily active users: ~100; daily active users

GeminiLLM EvaluationConversation AnalyticsVertex AI

Case Study

run_7f3a refund_agent Paused for approval

✓

lookup_order()

order A-1182 · delivered Oct 9
✓

check_refund_policy()

eligible · within 30 days
3

issue_refund()

amount: $120.00 → card ending 4417
4

send_confirmation()

waiting on previous step

state saved · agent worker released · resumes on decision

Approval request

Refund $120.00 to customer?

Toolissue_refund

Why pausedRefunds over $50 need review

Agent reasonItem arrived damaged

Approve Edit amount Reject

Sent to #support-approvals · expires in 4h

ADK Human-in-the-Loop Middleware

Personal

A pause-and-approve middleware for Google ADK agentic flows — open-source proof of concept.

Google ADKPythonAgentic SystemsHITL

GitHub Read on Medium

More work

Fedena Mobile App

Enterprise

0→1 launch of the Fedena Mobile App (Android + iOS) for school admins, parents, students, and teachers — drove a 30% adoption increase.

Gradebook · Exam structure Academic year 2019–20 ▾ Grade 6–8 ▾ + Add assessment

Scholastic assessment plan

Term 1 50% of final

Periodic Test 1

10%
Notebook Submission

5%
Subject Enrichment

5%
Half-Yearly Exam

80%

Term 2 50% of final

Periodic Test 2

10%
Notebook Submission

5%
Subject Enrichment

5%
Annual Exam

80%

Rules: periodic tests use the best of two attempts · marks rounded to the nearest whole number

Grading scale · 8-point

Co-scholastic · 3-point (A–C)

Work Education A
Art Education A
Health & Physical Ed. B
Discipline A

Fedena Gradebook

Enterprise

Designed and launched the Gradebook module across 10+ countries — 20% retention lift, 15% sales growth.

UniVault Link Manager

Personal

iOS link manager app — save, organize, and revisit URLs with smart tagging.

Show 4 more

Student ID template

Fields pulled directly from Fedena records

Photo
Full name
Admission no.
Class & section
Blood group
Guardian phone
QR code (student ID)

Greenfield Public School

Student Identity Card · 2020–21

Ananya Rao

Adm. no GPS-2041

Class 5 · B

Blood O+

Valid till
31 Mar 2021

Live preview · front

Print queue

Replacement ID cards

3 students · Grade 5

Printing
New staff ID cards

2 employees

Ready
Sports Day certificates

12 students · print + PDF

Queued

Print in-house in any quantity, from one card up

Fedena ID Cards

Enterprise

Let schools design and print their own student and staff ID cards and certificates from Fedena records, instead of relying on outside print vendors.

CX Agent Studio Product Review

Personal

Independent product breakdown of Google CX Agent Studio — connecting enterprise patterns to ADK HITL concepts.

Customizable Wordle App

Personal

A Flutter clone of Wordle with custom word lists, difficulty modes, and themes.

Custom Fitbit Watchface

Personal

A minimalist watchface built for the Fitbit SDK — daily goals, weather, and battery at a glance.

Stack & Approach

How I work

I'm a builder-first PM — comfortable in Figma, readable in code, and at home in the terminal. I've shipped across conversational AI, mobile apps, SaaS platforms, and data tools, and I bring the same structured thinking to every layer of the stack.

Tools & technologies

Google ADKVertex AIConversational AgentsCX Agent StudioDialogflowGemini APIsFlutterSwift / iOSFirebaseReactAstroTailwind CSSFigmaLookerMixpanelSQLPythonGitHubJiraNotion

My approach

▹ 0 → 1 product development — from problem discovery to shipped product
▹ Roadmap & backlog prioritization across competing stakeholder needs
▹ Cross-functional leadership with eng, design, data, and business teams
▹ Agile / Scrum delivery (CSPO certified) — sprints, retros, and release planning
▹ User research & discovery — interviews, journey mapping, and usability testing
▹ Agentic & conversational AI system design — flows, fallbacks, and HITL patterns
▹ Data-informed decisions using analytics, funnels, and experiment readouts
▹ Technical fluency across web, mobile, APIs, and cloud — enough to build prototypes