Headline

AI Product & AI Operations | Building toward AI Product Manager | I ship production AI bots enterprises trust | Conversational AI · LLMs · Multi-Agent Systems · Prompt Architecture · QA @ Kapture CX | KapHacks 5 🏆

About

Hey LinkedIn people 👋 I ship production AI bots that enterprises in BFSI, e-commerce, and microfinance actually trust — from prompt architecture to the QA systems that prove they work. At Kapture CX I own AI delivery across live conversational AI products: WhatsApp bots, voice bots, and multi-agent LLM systems handling real customer interactions at scale. What that's looked like: * Designed multi-agent architectures (orchestrator + specialist agents) with region-specific logic and guardrails against bad AI behavior, iterating prompts through 12+ versions off live session data * Built a UAT automation system — a FastAPI dashboard with Playwright browser automation, an LLM-as-judge evaluator, and a 73-case test suite — and presented it to our CPO and CEO * Built Python integrations wiring bots to live logistics APIs and CRM systems, debugging production issues under deadline pressure * Built the benchmarking framework for a Hindi/Hinglish voice bot, scoring intent accuracy, latency, and factual correctness against India-specific sources * Won KapHacks 5 🏆 - built a prompt library, now shipped in the Kapture product, that configures voice-bot prompts by vertical (BFSI, EdTech, and more) I sit where product judgment meets technical execution — deciding how a bot should behave, building the systems that make it behave that way, and proving it with data. Open to Technical Program/Project Manager (AI) and AI Product Manager roles — where technical depth and measurable outcomes both matter.

Location

Bengaluru

AI stack

antigravity

claude code

claude

cursor

gemini

github copilot

Work experience

Associate Project Manager - AI Delivery at Kapture CX, Apr 2026 - Present · 7 mos, Own end-to-end delivery of conversational AI products for enterprise clients across BFSI, e-commerce, and microfinance — from prompt architecture through production QA. - Designed and version-controlled multi-agent chatbot architectures (orchestrator + specialist sub-agents) for WhatsApp bots handling refunds, order status, and shipment tracking across multiple enterprise clients, including region-specific eligibility logic and guardrails against unreliable AI behavior - Built Python-based tool integrations connecting bots to live third-party APIs (shipment/logistics tracking) and internal CRM ticketing systems for a cross-border logistics client, including status-code mapping logic and a Bahasa Indonesia orchestrator prompt - Built and led a benchmarking framework for an outbound Hindi/Hinglish voice bot, scoring performance on call intent accuracy, latency, transcription mirroring, and factual accuracy against India-specific sources - Built and maintained a QA scoring system evaluating bot empathy and contextual response quality across hundreds of staging conversations, surfacing failure patterns before they reached production - Debugged and resolved production issues across the bot stack — including a webhook error, JavaScript formatting bugs, and authentication/config mismatches between staging and production environments - Audited and corrected 19 client-facing configuration touchpoints, improving reliability of client chat entry points - Queried and analyzed platform audit logs to debug client-specific traffic issues, establishing more reliable logging practices for the delivery team

Management Trainee - Ai-Ops at Kapture CX, Jun 2025 - Apr 2026 · 11 mos, Executed AI operations and conversational AI delivery across multiple enterprise clients, building foundational prompt architecture, solution documentation, and QA benchmarking systems for chatbot and voice bot products. - Built and delivered end-to-end solution documentation for a multi-flow WhatsApp chatbot, covering seven distinct conversational flows from client requirement to technical specification - Established foundational prompt engineering conventions for a production chatbot — including plain-text WhatsApp formatting rules, banned-phrase guardrails, and a region-detection priority chain — later scaled across multiple client deployments - Debugged and corrected backend integration functions (ticket creation/update), resolving authentication, endpoint, and payload mismatches between staging and production environments - Built and iterated a Hindi/Hinglish outbound voice bot, fixing language-detection logic, call-flow bugs, and script-generation rules to improve conversational reliability - Produced client-facing QA benchmark reports scoring voice bot performance on intent accuracy, latency, and factual correctness against India-specific data sources - Restructured and analyzed a client ticket dataset to identify root-cause patterns, presenting findings via a structured summary report to leadership - Built a project tracker mapping 30+ delivery tasks across 5 phases for a multi-workflow client onboarding, used to track cross-functional execution - Prototyped a behavioral AI concept ("Customer Twin") for a leadership-facing demo, including a working dashboard interface and backend architecture — presented to CTO-level stakeholders

Business Analyst at The White Mulberry, Feb 2025 - Jun 2025 · 5 mos, Noida, Uttar Pradesh, India

Deep Learning Engineer at AI4M Technology Private Limited, Jun 2024 - Dec 2024 · 7 mos, Pune District, Maharashtra, India, I am excited to start my new role as a Deep Learning Engineer Intern at AI 4M Technology, where I'll be working on cutting-edge AI projects and gaining and analysing hands-on experience in advanced machine learning techniques.

Deep Learning Engineer at AI4M Technology Pvt. Ltd., Jun 2024 - Dec 2024 · 7 mos, •​ Defined success metrics and problem scope for a manufacturing forecasting product; improved accuracy by 20% and reduced data pipeline processing time by 30%​

Co-Convener at Coding Club SRMAP, Aug 2023 - Jun 2024 · 11 mos

Education

Bachelor's degree, Computer Software Engineering, SRM University, AP

kendriya vidyalaya

Public links

github: https://github.com/1412naman