Headline

Co-founder & CTO @ Prodloop | Data & AI Engineering Leader | Agentic AI, LLM Evaluation, Voice AI | 13+ yrs in Data Architecture, MLOps & Multi-cloud | Ex-Accenture Applied AI Hub, BRIDGEi2i

About

Production AI on a strong data foundation: 13+ years across data engineering, cloud architecture, MLOps, pre-sales and applied AI, now building generative, agentic and voice AI. As Co-founder & CTO of Prodloop, I've grown a bootstrapped voice AI quality and observability SaaS with ~90% retention, serving 15+ enterprises across mobility, logistics, edtech and e-commerce. Built at Prodloop → Multi-tenant audit platform: customers review call and bot audits, reports and agent performance, and accept or dispute each parameter as human-in-the-loop feedback → LLM-as-judge evals at 95% accuracy vs human review across 13+ Indian languages and 10M+ calls; batch-first design cut inference cost 50% → Self-serve onboarding agent: customers upload calls, set rubrics and iterate on audits in chat, then hand off for integration with zero Prodloop effort → prod.ai: natural-language Q&A over calls on Vertex AI Search ("which agents were rude last week?", "who asked for a discount above INR 5,000?"), grounded in the calls → Voice agent orchestrator across STT, TTS, LLM and telephony with VAD and latency tuning, live with customers; pre-fetches CRM and WhatsApp context for a customer 360 view → Self-hosted multiple STT models and engineered open TTS models such as Kokoro for Indian languages on AWS GPUs, avoiding closed providers' per-token and concurrency fees → Built a TTS cache that reuses generated speech instead of regenerating it, cutting generation cost Data engineering & cloud → Pipelines from telephony, CRMs and internal systems into BigQuery with tenant isolation, PII masking and India data residency → Led end-to-end data engineering for multiple enterprise clients, heading delivery teams and partnering with client cloud and solution architects on data lakes, lakehouses, warehouses and MDM across GCP, Azure and AWS → Streaming pipelines, real-time microservices at sub-300ms, data quality engines, audit and balance controls, and MLOps with drift detection Pre-sales & go-to-market → RFPs, solutioning and reusable accelerators (data quality, MLOps) at BRIDGEi2i, so projects start from proven frameworks, cutting time to value and driving upsell → Founder-led sales at Prodloop, renewals at higher prices; event booths that won customers; active in AI meetups I rose from Data Modeler to Director, Technology Labs at BRIDGEi2i, then ML Engineering Senior Manager at Accenture AI Hub post-acquisition, leading teams of 8–10. Part of NSRCEL (IIM Bangalore) and NASSCOM GenAI Foundry; I write here on evaluation, agent design, MCP and inference economics.

Location

Bengaluru

AI stack

antigravity

claude

langgraph

langsmith

hermesagent

huggingface

Work experience

Co-founder & CTO at Prodloop, Jun 2024 - Present · 2 yrs 4 mos, Prodloop is an agentic AI platform for voice operations: quality, evals and self-improvement for voice bots and human agents. Bootstrapped, with ~90% retention and 15+ enterprise customers across mobility, logistics, edtech and e-commerce. Platform - Architected the multi-tenant platform: pipelines pull calls and metadata from telephony, CRMs and internal systems, run multi-hop processing across closed and open-source LLMs, validate against systems of record, and write results to customer CRMs, clouds or our audit portal - Audit portal where customers review bot and human audits, reports and agent performance, and accept or dispute each parameter; disputes feed a learning loop for our models and evals - LLM-as-judge evals at 95%+ accuracy vs human review across 22 Indian languages and code-switching, 10M+ conversations Agents - prodQA: voice-to-insight scoring of quality and compliance, including prosody (empathy, rate of speech, tonality) - Simulation agent built on 10K+ hours of Indian conversational data (LLM and voice simulations) - Funnel agent (where and why conversions drop), A/B testing agent (humans vs bots) and custom SOP agents for company-specific checks - Self-serve onboarding agent: customers upload calls, set rubrics and iterate on audits in chat before integration, no Prodloop team needed - prod.ai: natural-language Q&A over call data on Vertex AI Search, grounded in the calls Voice AI - Voice agent orchestrator across STT, TTS, LLM and telephony with VAD and latency tuning, live in production; pre-fetches CRM and WhatsApp context for a customer 360 view - Self-hosted STT and Kokoro TTS for Indian languages on AWS A10G GPUs (419ms median TTFB at 30 streams), plus a TTS cache - Cut inference cost 50% with batch-first, serverless design Outcomes - 25% fewer overcharging drivers (ride-hailing), 20% lower support cost and 8% higher CSAT (logistics), 12% incremental revenue (edtech), 20% higher assisted-buying conversion (e-commerce)

Founder & CTO at LEGOAI Technologies, Apr 2023 - Apr 2024 · 1 yr 1 mo, Bengaluru, Natural language analytics over enterprise data: business users ask questions in plain English and get answers from their own databases. - Built a deterministic relationship-discovery algorithm that connects to customer databases, infers which tables join on which columns, and returns a coverage-based confidence score - Built the natural language layer on top: a RAG pipeline (chunking, embeddings, retrieval) with on-the-fly SQL generation and an evaluation loop for answer precision - Self-hosted Falcon, Llama and Mistral on 8× NVIDIA A100 80GB (Azure NDm A100 v4) in 2023, exposed as a SQL-generation endpoint, because enterprise customers would not let even schema metadata leave their environment - Owned product, architecture and customer conversations end to end as founder

Accenture S&C (AI Hub) - ML Engineering - VP/Senior Manager (Bridgei2i Acquisition) at Accenture AI, May 2022 - Apr 2023 · 1 yr, Bengaluru, Joined Accenture Strategy & Consulting's AI Hub through Accenture's acquisition of BRIDGEi2i. - Led ML engineering delivery across enterprise client accounts, owning solution architecture and technical delivery - Built trust with client stakeholder teams and project sponsors on the accounts I led, turning delivery credibility into upsell and expanded scope - Handled RFP responses and solutioning for AI, ML and data engineering opportunities, presenting to prospective clients - Presented the accelerators built at BRIDGEi2i (data quality framework, MLOps platform) to show faster time to value, with projects starting from proven frameworks instead of from zero - Carried those accelerators into AI Hub engagements as reusable building blocks for client delivery - Led and mentored ML and data engineering teams across cloud platforms

Director - Technology Labs at BRIDGEi2i Analytics Solutions, Apr 2022 - Apr 2022 · 1 mo, Bengaluru, Promoted to Director, Technology Labs in April 2022, a month before BRIDGEi2i's acquisition by Accenture; the role mapped to Senior Manager at Accenture. Technology Labs was BRIDGEi2i's group for reusable accelerators and pre-sales solutioning. - Worked on RFP responses and solution design, presenting our accelerators to prospective clients as technical differentiators - The accelerators (data quality framework, MLOps platform) let client projects start from proven frameworks instead of from zero, cutting time to value and contributing to won deals and upsell - Senior technology architect and consultant across multiple client engagements

Cloud Architect at BRIDGEi2i Analytics Solutions, Oct 2020 - Mar 2022 · 1 yr 6 mos, Bengaluru, - HPE: architected an API and data mesh over a Hortonworks/Cloudera estate, with multi-layer caching holding sub-300ms response times for HPE's customer-facing web portal. The architecture my team built was awarded at HPE Discover in the US, presented there by HPE's own team - Worked directly with HPE's CDO office as sole client-facing owner, leading a team of 15+; the trust built with stakeholders and the project sponsor led to upsell, more revenue for BRIDGEi2i and a larger team on the account - Clorox (Azure): technical architect leading the team on an end-to-end data transformation platform, a metadata-driven lakehouse on Azure Data Factory, Databricks, ADLS, Synapse, Azure DevOps and Power BI - Cholamandalam: delivered a similar end-to-end Azure data lake and warehouse for the financial services group - Senior technology architect across multiple other client engagements

Technical Lead - Data Engineering & MLOps at BRIDGEi2i Analytics Solutions, Mar 2019 - Oct 2020 · 1 yr 8 mos, Bengaluru, - Clorox (US): led a greenfield marketing analytics platform end to end for Clorox's marketing team, working from Bangalore. Ingested 40+ marketing sources, including YouTube, Google campaign analytics, Google Maps, Instagram and Facebook, computed clicks, impressions and campaign KPIs, and fed Power BI dashboards used by the marketing head and team - Built it fully serverless on GCP (Cloud Scheduler, Pub/Sub, Cloud Storage, Cloud Functions, BigQuery) at a time when most projects still ran on servers - Learned GCP from scratch on this project and earned the Google Cloud Professional Data Engineer certification along the way - Built reusable data quality and MLOps accelerators, later used across client projects and in pre-sales

Data Modeler at BRIDGEi2i Analytics Solutions, Mar 2018 - Mar 2019 · 1 y, Bengaluru, Joined the team building BRIDGEi2i's two products, BridgePersona and BridgeFunnel. Designed the data model for both products Led the 8-person data engineering team behind them, taking the work from requirements through technology benchmarking to delivery (Python/FastAPI, Postgres, Neo4j, AWS)

Senior Software Engineer at Wipro Limited, Dec 2017 - Mar 2018 · 4 mos, Bengaluru Area, India, - Senior data modeller for State Street, a global financial services client, owning data model design and specifications on the engagement

Software Analyst at Accenture, Aug 2015 - Oct 2017 · 2 y 2 mo, Bengaluru, Nordea Bank engagement, based in Bengaluru with on-site work in Denmark. Data modeller and architect on a customer master data management programme that unified disconnected banking systems across Sweden, Finland, Norway and Denmark into a single customer 360 view Implemented the IBM MDM solution from scratch, acting as MDM product consultant across the end-to-end solution Owned the data model, the architecture and the specifications handed to the ETL team Monitored DataStage jobs pulling from country-specific mainframe sources into IBM MDM on DB2 Strong in OLTP and OLAP data modelling, SQL, data profiling and data quality

Senior Systems Engineer at IGATE, Apr 2013 - Aug 2015 · 2 y 4 mo, Mumbai, Data modeller for MetLife (US), designing OLTP and OLAP data models in ERwin Worked across data modelling, DataStage ETL and PL/SQL development ETL developer for Sydney Trains (Australian government), building DataStage jobs from multiple source systems into the reporting database, with Cognos reporting on top PL/SQL development for data warehouse workloads

Education

Bachelor of Technology - BTech, Computer Science and Engineering, Maharashtra Institute of Technology Aurangabad