📍 Udaipur, India  •  🎓 IIT Bhilai — B.Tech. Data Science & AI, 2028  •  🏆 Codeforces Specialist

🧠 About Me

engineer:
  name: "Vidit Shrimali"
  role: "AI Engineer & Competitive Programmer"
  focus: ["Machine Learning", "Multi-Agent Systems", "Real-world Deployment"]
  philosophy: "Bottom-up tinkering > Top-down theories"
  currently: "B.Tech in Data Science & AI @ IIT Bhilai (CGPA: 8.97/10)"

I build AI systems that leave the lab and actually ship — from a computer-vision pipeline deployed across 54 government health centres, to multi-agent voice systems making real outbound calls, to autonomous trading agents running 24/7. I compete on Codeforces to stay sharp on algorithms, and apply the same rigor to production ML.

💼 Experience

Machine Learning Engineer Intern — Anveo Innovations

Aug 2025 – Dec 2025 · Chhattisgarh, India

Built and deployed an AI-powered child growth assessment system for the District Collector of Dhamtari — automated height & weight estimation directly from images, replacing manual anthropometric measurement.

• Collected and curated data from 8,568 children across 258 centres, deployed across 54 centres
• Built an end-to-end inference pipeline (body segmentation → regression) served via AWS for scalable field use

🚀 Featured Projects

🔹 Velorah — Competitive Programming Intelligence Platform

Analyzed 100,000+ Codeforces records to uncover the real drivers of rating progression, then turned the findings into an interactive analytics platform.

• Feature-engineered log-linear regression models on contest activity & consistency metrics
• 10 interactive visualizations delivering data-driven improvement insights

🔹 Cold Calling Army — Multi-Agent Customer Acquisition Platform

A Hindi-first multi-agent voice system that autonomously qualifies leads and syncs to CRM in real time using the Gemini Live APIs.

• Production-grade agent orchestration on GCP with real-time tool use and observability
• Validated live in the field, not just in simulation

🔹 AlphaMemoir — Memory-Native Multi-Agent Trading System

A self-improving, 24/7 crypto trading platform (ccxt/Binance) orchestrating 8 LLM agents across a LangGraph pipeline with code-enforced risk guardrails.

• Retrieval-augmented memory layer on self-hosted Supermemory — 5 memory types, 768-dim ONNX vectors, hybrid semantic search
• Full-stack React dashboard backed by a 20-endpoint REST API

🛠️ Tech Stack

📫 Let's Connect