LIVE

◆ Neural Expertise Map

Live signal propagation across skill neurons

Pulin Pathneja

// Fractional CTO | Agentic AI & Data Platform Leader

20+ years designing, deploying, and scaling autonomous multi-agent systems across telecom, SaaS, and travel tech. Shipped production agentic AI: LLM-powered agents, self-optimizing Martech, predictive operations, serving 350M+ users on a 3.2T daily record data platform. Pioneered multi-agent orchestration with tool-use, RAG, and human-in-the-loop guardrails.

Multi-Agent Orchestration LLM Agents RAG Pipelines Tool-Use Human-in-the-Loop

✉ Contact Channel

Gurugram, Haryana

▲ System Metrics

Users Served

Telecom customer base

Daily Records

Real-time pipeline

AI Cost Savings

Energy anomaly detection

Engineers Led

Across 4 verticals

◆ System Analytics

Impact KPIs

89%Resolution Accuracy

80%Annotation Automation

+15%Engagement Lift

10xFaster Resolution

Agent Throughput (TPS)

Martech Platform

100K+

Samwad Agent

12K

NCH RCA Agent

Geo-Intel Engine

25K

Demand Gain

Data Volume Growth (Trillion Records/Day)

201720182019202020212022202320242025

Team Verticals

CLM Martech 35%

Location Intel 25%

Network DL 22%

Data Science 18%

◆ Skill Radar

Core Competency Map

Proficiency Breakdown

Agentic AI

95%

Data Platforms

92%

Cloud Infra

85%

Team Leadership

88%

Product Strategy

82%

GenAI / LLM

90%

■ Live Indicators

Campaign Throughput +12%

100K+ TPS

Agent Accuracy 89%

89% NCH

Data Processing 3.2T/d

3.2T records/day

Churn Reduction -10%

-10% HV users

■ Career Activity Heatmap 20 years

Contribution Intensity by Year

Less More

⚙ Active Agents 7 deployed

Agentic Martech Platform deployed

Telecom Enterprise

Autonomous campaign engine. Planner agents decompose goals, executor agents select channels & audiences, evaluator agents measure outcomes. LLM-driven real-time optimization across SMS, WhatsApp, App, Web.

100K+

TPS

<60ms

Latency

+15%

Engagement

-10%

Churn

Samwad & Saarthi deployed

Telecom Enterprise

Production LLM agents with tool-use, RAG pipelines, and structured output. Natural language interface to 350M+ user telecom datasets.

350M+users

RAG+ tool-use

NCH RCA Agent active

Telecom Enterprise

Autonomous root-cause analysis for network complaints. ML-driven diagnosis without human intervention.

30→3dTAT

89%accuracy

Geo-Intelligence Engine active

Telecom • DMRC

5+ TB daily geo-temporal data. Metro planning, AdTech, Digital Twin of India.

5+ TBdaily

DMRCgovt

Energy Anomaly Agent monitoring

Telecom Enterprise

PAN-India energy billing anomaly detection at national scale.

₹100Cr+saved

Annotation Agents advisory

Labellerr

Pre-labeling ML + QA agents. Confidence-based human handoff. 80% automation.

80%automation

CV+NLPpipelines

Demand Gain deployed

RateGain Travel

Autonomous dynamic pricing for travel. Multi-signal real-time rate optimization.

Real-timepricing

Multisignal

☰ Activity Log

Career Timeline 2005 – present

2020–2025

AVP, Big Data and AI

Telecom Enterprise

Enterprise agentic AI. Multi-agent systems, LLM agents, 3.2T records/day. 45+ engineers, 4 verticals.

2018–2020

Advisor

Labellerr

Autonomous annotation agents, 80% automation. Agent-human handoff protocols.

2017–2020

Head of Data & AI

RateGain Travel Technologies

Autonomous pricing & sentiment agents. DLaaS. GCP serverless migration.

2012–2017

Senior Consultant

Guavus

Spark, YARN, HBase. Telecom data pipelines at scale.

2005–2012

Software Engineer → Senior Consultant

Xebia • RBS • Oracle • Tavant

Enterprise foundations: MDM, mainframe migration, BI, CMS.

☍ Social Feed LinkedIn

Pulin Pathneja

Agentic AI Leader • 1d

Everyone's talking about agentic AI.

Almost no one is actually deploying it.

Gartner just reported a 1,445% surge in multi-agent system inquiries. But Deloitte shows only 11% are running agents in production.

That's a 130x gap between interest and execution. Why?

Because we're obsessing over the wrong thing. The model is 5% of the system. The orchestration, guardrails, tool integration, and observability. That's where companies die.

The trough of disillusionment isn't a death sentence. It's a filter. It separates demo builders from system builders.

#AgenticAI #MultiAgentSystems #LLM #MCP

👍 142 💬 38 comments 🔄 24 reposts

Pulin Pathneja

Agentic AI Leader • 5d

5 years of deploying AI agents at 350M-user scale. Here's what most people get wrong:

“Build a smarter agent” → Wrong.
Build a smarter system of agents.

Our Martech platform ran 3 agents in a loop:

• Planner (decomposes goals)
• Executor (acts on channels)
• Evaluator (measures + feeds back)

100K+ TPS. Sub-60ms. 15% more engagement.

The agentic future isn't about one brilliant agent. It's about how agents work together.

#AgenticAI #AI #DataPlatform

👍 89 💬 21 comments 🔄 15 reposts

Pulin Pathneja

Agentic AI Leader • 2w

We reduced network complaint resolution from 30 days to 3 days at scale. Not by hiring more people. By deploying autonomous RCA agents.

The agents diagnose root causes across millions of network signals, prioritize by severity, and trigger resolution workflows, all without human intervention.

Accuracy went from 65% to 89%.

The best AI isn't the one that generates text. It's the one that solves problems while you sleep.

#AI #Telecom #PredictiveOps #Automation

👍 204 💬 47 comments 🔄 31 reposts

★ System Capabilities

Agentic AI

Multi-Agent Orchestration LLM Agents in Production Tool-Use RAG Pipelines Human-in-the-Loop Autonomous Workflows Agent Guardrails Agentic Martech

AI & Data Leadership

AI Strategy MLOps / LLMOps GenAI NLP Computer Vision Data Monetization Product Strategy

Infrastructure

Python GCP AWS BigQuery Kafka Spark Airflow Docker K8s LangChain Vector DBs

Education

BE, Information Technology

Punjab Engineering College, Chandigarh

2001 – 2005

Open for new missions.

Fractional CTO available for agentic AI strategy, multi-agent system architecture, and engineering leadership in high-growth, product-led organizations.

Email Phone LinkedIn