Who I Am
AI Software
Engineer
I work at *astTECS building production-grade voice AI systems that handle real calls, real users, and real scale. I also co-founded Voxels Digital Agency, where we help businesses integrate AI into their operations.
My focus is on building systems that actually work in production: voice agents, LLM pipelines, computer vision models, and the infrastructure that holds it all together.
Education
Where I
Studied
A B.E. in AI/ML from Bangalore, a Stanford Machine Learning Specialization, and the science foundation that started it all.
B.E. in Artificial Intelligence & Machine Learning
AMC Engineering College, Bangalore
CGPA: 7.97/10
Machine Learning Specialization
Stanford University (Coursera)
Supervised Learning, Advanced Algorithms, Unsupervised Learning
Pre-University (PCMB)
St. Joseph's Indian Composite PU College
Physics, Chemistry, Mathematics, Biology
Career
Journey
Three roles. Each one building on the last.
AI Software Engineer
*astTECS Unified Communication PVT LTD
Architected a production-grade multi-language voicebot handling 90 concurrent calls, achieving 40-50% workforce reduction. Built India's first MCP server integration with IP-PBX for agentic AI announcements. Designed LangChain agents as the core decision layer for multiple systems, integrated with SIP servers, and deployed self-hosted Langfuse for per-call observability and evaluation.
AI Trainer
Sambhav Foundation
Designed an AI/ML curriculum and mentored aspiring engineers. Covered supervised learning, neural networks, and practical model training. Made complex concepts approachable for students with no prior background in the field.
AI Internship
Codsoft
Built foundational ML projects including classification models and data pipelines. First hands-on exposure to production deployment workflows and collaborative software development.
Career
Circuit
AI Software Engineer
*astTECS Unified Communication PVT LTD
Architected a production-grade multi-language voicebot handling 90 concurrent calls, achieving 40-50% workforce reduction. Built India's first MCP server integration with IP-PBX for agentic AI announcements. Designed LangChain agents as the core decision layer for multiple systems, integrated with SIP servers, and deployed self-hosted Langfuse for per-call observability and evaluation.
AI Trainer
Sambhav Foundation
Designed an AI/ML curriculum and mentored aspiring engineers. Covered supervised learning, neural networks, and practical model training. Made complex concepts approachable for students with no prior background in the field.
AI Internship
Codsoft
Built foundational ML projects including classification models and data pipelines. First hands-on exposure to production deployment workflows and collaborative software development.
Impact & Recognition
Real-World
Results
Measurable results from production deployments. Numbers that came from real systems, not demos.
Star Performer
AI Team — *astTECS Communication Pvt. Ltd.
Recognized for exceptional contributions to production AI systems — including building a 90-channel multilingual voicebot, pioneering India's first MCP + IP-PBX integration, and delivering 40-50% workforce optimization through intelligent automation.
View in astTECS Annual Magazine 2026Concurrent Voice Capacity
Production-grade multilingual voicebot handling 90 concurrent calls with real-time STT/TTS processing on completely local infrastructure.
Workforce Optimization
Delivered measurable business impact through intelligent automation, reducing manual workload by 40-50% while improving response times.
MCP + IP-PBX Integration
Pioneered India's first Model Context Protocol server integration with IP-PBX systems for agentic AI voice announcements.
Featured in TimesTechStar Performer — AI Team
Recognized as Star Performer in the AI Team for exceptional contributions to production AI systems, innovation in voice AI infrastructure, and consistent delivery of high-impact solutions.
Featured in astTECS Annual Magazine 2026 — Pg 17Production System
Enterprise
Voice AI
A full-stack agentic telephony platform running in production at *astTECS.
Replaced a state-machine + Rasa-based PBX system with a fully agentic voice AI built on the LiveKit Agents framework. The agent runs on self-hosted LiveKit WebRTC with a SIP trunk, handling inbound calls, routing to extensions via SIP REFER, scheduling callbacks, and running live web searches.
Every call is recorded via LiveKit Egress, transcribed turn-by-turn, and processed post-call by a single LLM inference that extracts sentiment, disposition, intent, and an AI summary. Authorized callers can query CDR reports by department or individual, with results sent over SMTP. The dashboard surfaces the full call log with per-entry WAV recordings and transcripts.
STT runs on self-hosted NVIDIA Parakeet. TTS runs on self-hosted Svara. All traces go to self-hosted Langfuse with per-call spans covering every pipeline stage.
377
Total calls processed
61%
Transfer success rate
51s
Avg handle time
Metrics as of Apr 14, 2026 — 1 week into production.
Internal Transfer
SIP REFER via LiveKit, 61% success rate
CDR Reports
By dept / employee, emailed on demand
Post-Call Analysis
Sentiment, disposition, intent, AI summary — single LLM call
Fully Self-Hosted
LiveKit, SIP, Egress, LLM, Parakeet STT, Svara TTS
Observability
Langfuse traces per call — STT, LLM, tool, TTS spans
Caller Auth
Extension-level access control for CDR queries
System Output
Live Calls
Three real interactions handled by the agent — click any row to see the recording, transcript, and AI post-call analysis.
Technical Arsenal
Tools of the
Machine
AI & ML
Data Science
Tools & Frameworks
Selected Works
Conversational AI
Real-Time Voice AI Agent
Machine Learning & Finance
Stock Price Prediction
Computer Vision
Dog Vision
Multi-Modal AI
VideoSearch AI
Generative AI
Personalised Image Generation
Healthcare AI
Heart Disease Classifier
Start a Conversation
Let's Work
Together
Whether you need AI integration, custom ML models, or technical consultation, reach out and I'll get back to you.