Headline

AI/ML | Python, SQL, Agentic AI, RAG, GNNs | Building Intelligent Systems with LLMs & Machine Learning | Open to Entry Level Roles

About

AI/ML Engineer specializing in Generative AI, Agentic AI, and Voice AI systems, with hands-on experience building and deploying production-grade AI solutions for real-world clients. Currently a Forward Deployed Engineer Intern at Agentic Universe, configuring voice AI agents for BFSI clients handling 25K–30K+ outbound calls/day, working directly with clients to improve agent performance — contributing to a 50–60% lift in lead-to-conversion rates. Previously built an Industrial AI QualityOps Copilot (Detectron2, CAD-based defect detection) as an AIML Intern at Zero Zeta. Project experience: RAG systems (LlamaIndex, Qdrant), multi-agent workflows (LangGraph), and Graph Neural Networks for cybersecurity (98% recall, 300K+ records). Skills: Python, PyTorch, LangGraph, LlamaIndex, RAG, Agentic AI, Voice AI (TTS/STT), Computer Vision, GNNs, Qdrant, Supabase, REST APIs.

Location

Bengaluru

AI stack

aistudio

claude code

codex

cursor

mcp

openai

Work experience

Forward Deployed Engineer at Agentic Universe, Sep 2026 - Present · 1 mo, Bengaluru, • Configure and maintain production voice AI agents handling 25K-30K+ outbound calls per day per client for BFSI clients, owning agent setup, conversation flow design, and call-handling parameters at scale • Act as the client-facing technical contact across accounts, translating evolving client expectations into configuration and prompt updates that lifted lead-to-conversion rates by 50% • Developed an end-to-end understanding of the voice backend system, including how the STT, LLM, and TTS layers work together to deliver live calls • Monitor system-level agent performance using AWS CloudWatch logs and metrics to catch anomalies before they affect calls • Diagnose live issues and perform root cause analysis (RCA) backed by CloudWatch log evidence, so clients get clear, proof-based explanations of what went wrong and how it was fixed

AIML Intern at Zero Zeta, Jul 2026 - Sep 2026 · 3 mos, Remote, • Developed an Industrial AI QualityOps Copilot that extracts dimensions/tolerances from DXF/DWG CAD files and uses Detectron2(ResNet-50 + Mask R-CNN) for component detection and contour-based geometric measurement from camera images. • Built an end-to-end CAD-to-image measurement and tolerance-validation pipeline, using calibrated pixel-to-mm conversion and automated feature mapping to generate dimension-level PASS/FAIL assessments, and deployed the solution as a production web application through direct client collaboration

Education

Bachelor of Engineering - BE, Artificial intelligence and machine learning, nitte meenakshi institute of technology

12, science, kendriya vidyalaya

Public links

github: https://github.com/AryanPandey72