Looking for a job

Vasu Verma

@vasuverma

Message

I build production voice AI, LLM fine-tuning platforms, and multi-agent systems for enterprise workflows.

India

Message

What I'm looking for

I'm looking to build production GenAI products, especially voice AI, LLM platforms, and agentic systems, where I can own architecture and product outcomes.

At Metafore, I founded and own Mira, a real-time conversational voice AI agent for telco customer care and second-line support. I lead its product scope, architecture, GTM narrative, pricing, and enterprise pitches.

I architected Mira from SIP ingress and telecom infrastructure through LLM orchestration, enabling direct carrier-voice integration. I reduced live-call turn latency from roughly 500 ms to under 300 ms and built a 50+ tool multi-agent system for CRM, billing, plan changes, ticketing, and network diagnostics.

I also shipped an AWS-based LLM fine-tuning platform adopted by 15+ engineers, with LoRA/QLoRA workflows, DVC and MLflow versioning, containerized inference, GPU autoscaling, and monitoring. Previously at LSEG, I automated Salesforce UAT validation and built LLM-driven reconciliation across 12 Cr-record datasets at 93% accuracy.

My earlier work spans FastAPI and Neo4j knowledge-graph services at Siemens, multilingual healthcare RAG at SmartQR, and audio ML at boAt. I enjoy taking GenAI systems from model experimentation to production products with measurable business outcomes.

Experience

Work history, roles, and key accomplishments

Current

Full Stack Gen AI Research Engineer

Current

Metafore

Jan 2026 - Present (8 months)

Initiated and own Mira, a real-time voice AI agent for telco customer care, architected from SIP ingress to LLM orchestration, and reduced latency to under 300 ms. Built a multi-agent system with 50+ tools and delivered POCs with Airtel, Orange, and Axiata.

Generative AI & ML Intern

London Stock Exchange Group (LSEG)

May 2025 - Jan 2026 (8 months)

Automated Jira-story-to-UAT Salesforce validation using Python and GenAI orchestration, reducing engineer effort by 97%. Engineered LLM-driven reconciliation across 12 Cr-record datasets, detecting mismatches at 93% accuracy.

Python GenAI LLM Salesforce Jira UAT QA Data Reconciliation

GenAI & Knowledge Graph Intern

Siemens Technology and Services

May 2024 - Jul 2024 (2 months)

Built a FastAPI service for dual-PDF protocol mapping, auto-generating Neo4j knowledge graphs from embeddings with PMI-weighted edges. Designed an NLP-to-SHACL pipeline validated via pySHACL for domain-agnostic RDF generation.

fastAPI Neo4j Cypher Knowledge Graphs NLP SHACL Resource Description Framework (RDF)LLM Streamlit Python

Lead GenAI Developer

SmartQR

Dec 2023 - Mar 2024 (3 months)

Deployed a 15-language healthcare chatbot for India's Home-to-Home medical program, achieving 90% on F1/precision/BLEU. Built a women's and adolescent-health LLM outperforming GPT-3 by 5% on clinical response accuracy.

LLaMA Hugging Face RAG Python Image Processing Healthcare LLM BLEU

Embedded Development Intern

boAt

May 2023 - Jun 2023 (1 month)

Developed an Enhanced Noise Cancellation algorithm achieving 83% improvement and trained CNNs on 1L MFCC audio samples at 93% accuracy.

C C++Embedded Systems Signal Processing CNN MFCC (Feature Extraction)Audio Processing

Education

Degrees, certifications, and relevant coursework

Indian Institute of Technology Kharagpur

Bachelor of Science, Geology & Geophysics

Grade: 7.70 / 10

B.S. in Geology & Geophysics with a micro-specialization in Entrepreneurship & Innovation Engineering.

Amazon

Amazon ML Summer School, Machine Learning

2024 - 2024

Selected among top 0.6% nationwide; trained by Amazon Scientists in machine learning topics including supervised learning, deep neural networks, and generative AI.