Summary
Overview
Work History
Education
Skills
Certification
Languages
Accomplishments
Affiliations
Timeline
Generic
TARUN TYAGI

TARUN TYAGI

Summary

Dynamic Technical Lead with a proven track record at Truminds Software Systems, specializing in AI/ML platforms and multi-agent orchestration. Achieved a 30% reduction in manual work through innovative solutions. Expert in LLM integrations and passionate about driving automation and efficiency in enterprise operations while fostering collaborative team environments.

Overview

6
6
years of professional experience
1
1
Certification

Work History

Technical Lead - AI/ML

Truminds Software Systems
Gurugram
03.2022 - Current
  • Architected and led development of TruAgent, a multi-agent, no-code AI platform enabling enterprises to create domain-specific intelligent assistants using LLMs and agentic orchestration.
  • Designed a modular microservices architecture supporting plug-and-play AI agents, MCP integrations, voice and avatar modules, ensuring high scalability and modularity.
  • Integrated multiple LLM and embedding models (OpenAI, Anthropic, Ollama, and custom MCP servers) for configurable model routing and context adaptation.
  • Implemented advanced Retrieval-Augmented Generation (RAG) pipelines using LangChain and LangGraph, supporting multi-source ingestion (websites, documents, Confluence, cloud storages).
  • Enhanced semantic retrieval through Neo4j knowledge graphs, improving response accuracy and contextual understanding by 35%.
  • Developed specialized AI agents—Database Agent, RAG Agent, Visualization Agent, and Analytics Agent—enabling data-driven automation across enterprise operations.
  • Integrated LangFuse for AI observability, tracking token usage, latency, and prompt performance with feedback loops for model improvement.
  • Implemented GuardRails AI for prompt validation, compliance, and data security, ensuring safe and reliable AI interactions.
  • Engineered real-time voice conversation pipelines using LiveKit (WebRTC) for low-latency communication and ElevenLabs for lifelike speech synthesis, enabling bi-directional conversational AI.
  • Integrated Live avatars using Hedra for immersive multimodal interactions, combining LLM intelligence with real-time audio-video engagement.
  • Delivered measurable outcomes: 30% reduction in manual work, 25% operational cost savings, and 2× faster query resolution across enterprise workflows.
  • Built in-platform analytics dashboards to monitor agent effectiveness, user engagement, and model performance metrics in real time.
  • Led automation projects using Python, PyATS, and Ansible, creating automated pipelines for 5G NF configuration backup, restore, and regression testing—reducing QA effort by 40%.
  • Developed a custom Router OS based on OpenWrt, integrating features like captive portal, user management, VLAN CRUD, and firewall automation, improving network performance and control.
  • Designed real-time media streaming pipelines for RTSP cameras using HLS and WebRTC, optimizing latency for live monitoring and playback.
  • Collaborated across multi-domain client projects in AI/ML, networking, and edge computing, delivering scalable and production-grade systems.

Lead Data Scientist

CogniTensor Pvt Ltd
Noida
02.2020 - 03.2022
  • Built data and ML pipelines, reducing deployment errors by 35% and accelerating model release cycles.
  • Applied time series models (ARIMA, SARIMA, LSTM, N-BEATS, GluonTS) to forecast events with >90% prediction accuracy for disaster and sales planning.
  • Conducted comparative analysis of classical and deep learning models, improving forecasting efficiency by 15%.
  • Implemented statistical techniques (Mean, AR, Moving Average) to analyze trends and correlations across daily, weekly, and monthly intervals.
  • Developed recommender systems using collaborative filtering, increasing user engagement and click-through rates by 20%.
  • Built chatbots and voice assistants with Rasa Core, Rasa NLU automating >50% of customer interactions and reducing manual response time.

Education

Post-Graduation - Data Science and Engineering

Great Lakes Executive Learning
Gurgaon
01.2020

M.Tech + B.Tech - Nano Technology

Amity University
Noida, Uttar Pradesh
08.2018

Skills

  • AI/ML Platforms: LangChain, LangGraph, Neo4j, LangFuse, GuardRails AI, Weaviate — for multi-agent orchestration, RAG, and semantic retrieval
  • LLM Integrations: OpenAI, Anthropic, Ollama, Custom MCP Servers — expertise in fine-tuning, prompt engineering, and context optimization
  • Deep Learning & Quantization: TensorFlow, PyTorch, ONNX Runtime — model compression (INT8/FP16), pruning, and distillation for edge deployment
  • Classical Machine Learning: Regression, XGBoost, Random Forest, SVM, Clustering (K-Means, DBSCAN), and feature engineering with strong evaluation metrics (F1, ROC-AUC)
  • Conversational AI: LiveKit (WebRTC), ElevenLabs, HeyGen — real-time voice synthesis, avatar interactions, and multimodal conversational pipelines
  • Automation & Infrastructure: Python, Ansible, PyATS, Docker, Kubernetes, Redis — CI/CD, container orchestration, and system automation
  • Architecture & Systems Design: Microservices, Multi-Agent Systems, Event-Driven Design, Graph-Enhanced RAG, and modular plugin architecture
  • Observability & Security: LangFuse, Prometheus, Grafana, GuardRails AI — model monitoring, compliance, and AI safety enforcement
  • Cloud & MLOps: AWS (S3, EC2, SageMaker), Azure (Functions, Cognitive Services), MLflow — scalable deployment and lifecycle management of AI models
  • Networking & Edge Computing: OpenWrt, RTSP, HLS, WebRTC — optimized streaming, real-time analytics, and edge AI system design

Certification

Introduction to Artificial Intelligence, Great Learning, Great Lakes

Languages

Hindi
First Language
English
Proficient (C2)
C2

Accomplishments

  • Awarded as employee of the year, 04/01/20 - 03/31/21
  • Awarded as most improved performance of the quarter, 06/01/20 - 09/30/20

Affiliations

  • Presented a paper in 60TH DAE Solid State Physics Symposium, Noida
  • Exploring the Cr2+ Doping Effect on Structural, Vibrational and Dielectric Properties of Mn-Zn Ferrites, AIP Conf. Proc. 1731, 090008, P. Choudhary, Tarun Tyagi, M.A. Dar, Dr. Dinesh Varshney

Timeline

Technical Lead - AI/ML

Truminds Software Systems
03.2022 - Current

Lead Data Scientist

CogniTensor Pvt Ltd
02.2020 - 03.2022

Post-Graduation - Data Science and Engineering

Great Lakes Executive Learning

M.Tech + B.Tech - Nano Technology

Amity University

Similar Profiles

CREATE PROFILE
TARUN TYAGI